1 2017-11-09 Martin Liska <mliska@suse.cz>
4 * coverage.c (coverage_begin_function): Output also end locus
5 of a function and information whether the function is
7 * gcov-dump.c (tag_function): Parse and print the information.
8 * gcov.c (INCLUDE_MAP): Add include.
9 (INCLUDE_SET): Likewise.
10 (struct line_info): Move earlier in the source file because
11 of vector<line_info> in function_info structure.
12 (line_info::line_info): Likewise.
13 (line_info::has_block): Likewise.
14 (struct source_info): Add new member index.
15 (source_info::get_functions_at_location): New function.
16 (function_info::group_line_p): New function.
17 (output_intermediate_line): New function.
18 (output_intermediate_file): Use the mentioned function.
19 (struct function_start): New.
20 (struct function_start_pair_hash): Likewise.
21 (process_file): Add code that identifies group functions.
22 Assign lines either to global or function scope.
23 (generate_results): Skip artificial functions.
24 (find_source): Assign index for each source file.
25 (read_graph_file): Read new flag artificial and end_line.
26 (add_line_counts): Assign it either to global of function scope.
27 (accumulate_line_counts): Isolate core of the function to
28 accumulate_line_info and call it for both function and global
30 (accumulate_line_info): New function.
31 (output_line_beginning): Fix GNU coding style.
32 (print_source_line): New function.
33 (output_line_details): Likewise.
34 (output_function_details): Likewise.
35 (output_lines): Iterate both source (global) scope and function
37 (struct function_line_start_cmp): New class.
38 * doc/gcov.texi: Reflect changes in documentation.
40 2017-11-09 Jakub Jelinek <jakub@redhat.com>
43 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
44 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
45 and similarly for not instead of neg.
47 2017-11-08 Andi Kleen <ak@linux.intel.com>
49 * config/i386/i386.opt: Add -mforce-indirect-call.
50 * config/i386/predicates.md: Check for flag_force_indirect_call.
51 * doc/invoke.texi: Document -mforce-indirect-call
53 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
55 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
57 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
58 config/riscv/riscv.c (predict.h): New include.
59 (riscv_slow_unaligned_access_p): No longer static.
60 (riscv_block_move_straight): Add require.
61 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
63 2017-11-08 Jakub Jelinek <jakub@redhat.com>
66 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
67 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
68 nonimmediate_operand predicate for operand 1 instead of
71 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
73 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
75 * config/aarch64/constraints.md (Uml): New constraint.
76 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
79 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
81 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
82 of two vec_duplicates into a vec_concat.
84 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
86 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
87 Simplify vec_merge of vec_duplicate and vec_concat.
88 * config/aarch64/constraints.md (Utq): New constraint.
89 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
92 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
94 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
95 Simplify vec_merge of vec_duplicate and const_vector.
96 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
98 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
99 mode iterator. Update predicate on operand 1 to
100 handle non-const_vec constants. Delete constraints.
101 (*aarch64_combinez_be<mode>): Likewise for operand 2.
103 2017-11-08 Jakub Jelinek <jakub@redhat.com>
105 PR tree-optimization/78821
106 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
108 (store_operand_info::store_operand_info): Initialize it to false.
109 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
110 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
111 store in the group, and if chain_info is non-NULL, to ignore altogether
113 (compatible_load_p): Fail if bit_not_p does not match.
114 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
115 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
116 (pass_store_merging::process_store): Adjust
117 terminate_all_aliasing_chains calls to pass NULL in all current spots,
118 call terminate_all_aliasing_chains newly when adding a store into
119 a chain with non-NULL chain_info.
121 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
123 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
125 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
127 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
129 (aarch64_layout_frame): Initialise emit_frame_chain.
130 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
131 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
133 2017-11-08 Martin Liska <mliska@suse.cz>
135 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
136 of gimple_call_internal_p.
138 2017-11-07 Tom de Vries <tom@codesourcery.com>
140 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
142 2017-11-07 Tom de Vries <tom@codesourcery.com>
144 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
147 2017-11-08 Martin Liska <mliska@suse.cz>
150 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
152 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
154 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
156 2017-11-07 Jakub Jelinek <jakub@redhat.com>
159 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
160 (*cmp<mode>_ccz_1): New insn with $k alternative.
163 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
164 RTX_COMM_COMPARE as commutative as well.
165 (ix86_binary_operator_ok): Formatting fix.
166 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
167 *<code><mode>3<mask_name><round_saeonly_name>,
168 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
169 *<s>mul<mode>3_highpart<mask_name>,
170 *vec_widen_umult_even_v16si<mask_name>,
171 *vec_widen_umult_even_v8si<mask_name>,
172 *vec_widen_umult_even_v4si<mask_name>,
173 *vec_widen_smult_even_v16si<mask_name>,
174 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
175 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
176 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
177 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
178 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
179 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
180 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
181 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
182 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
183 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
184 ix86_binary_operator_ok. Formatting fixes.
185 (*<plusminus_insn><mode>3<mask_name><round_name>,
186 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
189 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
191 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
192 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
193 eq and ne if TARGET_ISEL.
194 (cmp): New code_iterator.
195 (UNS, UNSU_, UNSIK): New code_attrs.
196 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
197 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
199 ("eq<mode>3"): ... this.
200 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
202 ("ne<mode>3"): ... this.
204 2017-11-07 Julia Koval <julia.koval@intel.com>
207 * common/config/i386/i386-common.c
208 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
209 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
210 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
211 * config/i386/i386.opt: Ditto.
212 * config/i386/i386.c (ix86_target_string): Ditto.
213 (ix86_option_override_internal): Ditto.
214 (ix86_init_mpx_builtins): Move MPX to args2.
215 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
216 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
217 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
218 __builtin_ia32_vgf2p8affineinvqb_v32qi,
219 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
220 __builtin_ia32_vgf2p8affineinvqb_v16qi,
221 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
223 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
226 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
227 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
228 (zero-extendsidi peephole2): Remove peephole.
230 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
233 * stor-layout.c: Include attribs.h.
234 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
235 explicit lookup of "aligned" attribute.
237 2017-11-07 Andrew Waterman <andrew@sifive.com>
239 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
240 (riscv_expand_block_move): Likewise.
241 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
243 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
244 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
245 * config/riscv/riscv.c (riscv_block_move_straight): New function.
246 (riscv_adjust_block_mem): Likewise.
247 (riscv_block_move_loop): Likewise.
248 (riscv_expand_block_move): Likewise.
249 gcc/config/riscv/riscv.md (movmemsi): New pattern.
251 2017-11-07 Michael Clark <michaeljclark@mac.com>
253 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
254 (MUSL_DYNAMIC_LINKER): Likewise.
256 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
258 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
259 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
262 2017-11-07 Richard Biener <rguenther@suse.de>
264 * match.pd: Fix build.
266 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
267 Jackson Woodruff <jackson.woodruff@arm.com>
269 PR tree-optimization/71026
270 * match.pd: Canonicalize negate in division.
272 2017-11-07 Sudakshina Das <sudi.das@arm.com>
275 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
277 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
279 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
280 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
282 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
284 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
285 non-scalar integral types.
286 * match.pd (negate_expr_p): Handle MINUS_EXPR.
287 (-(A-B), -(~A)): New transformations.
289 2017-11-07 Tom de Vries <tom@codesourcery.com>
291 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
292 semicolon after "do {} while (0)".
293 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
294 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
295 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
296 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
297 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
299 2017-11-07 Tom de Vries <tom@codesourcery.com>
301 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
302 after "do {} while (0)".
303 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
304 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
305 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
306 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
307 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
309 2017-11-07 Tom de Vries <tom@codesourcery.com>
312 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
314 (arm_rtx_costs_internal): Add missing semicolon after
315 HANDLE_NARROW_SHIFT_ARITH call.
317 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
319 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
320 disable isel if it was not set explicitly.
322 2017-11-06 James Bowman <james.bowman@ftdichip.com>
324 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
326 (add_type_attribute) likewise.
328 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
330 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
331 of struct ix86_frame.
332 (ix86_initial_elimination_offset): Likewise.
333 (ix86_expand_split_stack_prologue): Likewise.
335 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
337 * tree-vrp.h (enum value_range_type): Update stale comment.
339 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
340 Alan Hayward <alan.hayward@arm.com>
341 David Sherwood <david.sherwood@arm.com>
343 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
344 (aarch64_expand_vec_perm_const): Take the number of units too.
345 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
346 (aarch64_expand_vec_perm_const): Likewise.
347 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
348 (vec_perm<mode>): Update accordingly.
350 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
351 Alan Hayward <alan.hayward@arm.com>
352 David Sherwood <david.sherwood@arm.com>
354 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
355 Take the number of units too.
356 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
357 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
358 but check for a vector mode before rather than after the call.
359 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
360 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
361 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
362 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
363 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
364 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
365 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
366 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
367 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
368 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
369 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
370 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
371 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
372 (aarch64_sqdmull2_n<mode>): Update accordingly.
374 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
375 Alan Hayward <alan.hayward@arm.com>
376 David Sherwood <david.sherwood@arm.com>
378 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
379 the number of units too.
380 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
381 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
382 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
383 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
384 (vec_store_lanesxi<mode>): Update accordingly.
386 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
387 Alan Hayward <alan.hayward@arm.com>
388 David Sherwood <david.sherwood@arm.com>
390 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
391 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
392 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
393 of units rather than the mode.
394 * config/aarch64/iterators.md (nunits): New mode attribute.
395 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
396 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
397 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
398 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
399 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
400 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
401 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
402 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
403 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
404 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
405 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
406 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
407 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
408 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
409 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
410 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
411 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
412 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
413 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
414 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
415 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
416 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
417 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
418 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
419 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
420 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
421 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
422 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
423 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
424 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
425 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
426 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
427 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
428 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
429 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
430 (aarch64_simd_vec_setv2di): Likewise.
432 2017-11-06 Carl Love <cel@us.ibm.com>
434 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
436 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
437 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
439 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
441 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
442 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
443 Add power 8 macro expansions.
444 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
445 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
446 power 8 instructions. (VSX_XXBR): Add iterator.
448 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
450 * config/arm/arm.md (predicable_short_it): Change default to "no",
451 improve documentation, remove uses that are identical to the default.
452 (enabled_for_depr_it): Rename to enabled_for_short_it.
453 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
454 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
455 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
456 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
457 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
459 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
462 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
463 float128 helper macros, which are no longer used after deleting
464 the old 'q' built-in functions, and moving the round to odd
465 built-in functions to being special built-in functions.
466 (BU_FLOAT128_2): Likewise.
467 (BU_FLOAT128_1_HW): Likewise.
468 (BU_FLOAT128_2_HW): Likewise.
469 (BU_FLOAT128_3_HW): Likewise.
470 (FABSQ): Delete old 'q' built-in functions.
471 (COPYSIGNQ): Likewise.
472 (SQRTF128_ODD): Move round to odd built-in functions to be
473 special built-in functions, so that we can handle
474 -mabi=ieeelongdouble.
475 (TRUNCF128_ODD): Likewise.
476 (ADDF128_ODD): Likewise.
477 (SUBF128_ODD): Likewise.
478 (MULF128_ODD): Likewise.
479 (DIVF128_ODD): Likewise.
480 (FMAF128_ODD): Likewise.
481 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
482 built-in names to 'f128'.
483 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
484 old 'q' built-in functions, as the machine independent code for
485 'f128' built-in functions handles this.
486 (rs6000_expand_builtin): Add expansion for float128 round to odd
487 functions, keying off on -mabi=ieeelongdouble of whether to use
488 the KFmode or TFmode variant.
489 (rs6000_init_builtins): Initialize the _Float128 round to odd
491 * doc/extend.texi (PowerPC Built-in Functions): Document the old
492 _Float128 'q' built-in functions are now mapped into the new
493 'f128' built-in functions.
495 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
497 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
498 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
500 2017-11-06 David Malcolm <dmalcolm@redhat.com>
503 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
504 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
505 * toplev.c: Include "ipa-fnsummary.h".
506 (toplev::finalize): Call ipa_fnsummary_c_finalize.
508 2017-11-06 Jakub Jelinek <jakub@redhat.com>
510 PR tree-optimization/82838
511 * gimple-ssa-store-merging.c
512 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
513 on a separate gimple_seq which is then appended to seq.
515 2017-11-06 Jeff Law <law@redhat.com>
518 * config/i386/i386.c (PROBE_INTERVAL): Remove.
519 (get_probe_interval): New functions.
520 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
521 (ix86_adjust_stack_and_probe): Likewise.
522 (output_adjust_stack_and_probe): Likewise.
523 (ix86_emit_probe_stack_range): Likewise.
524 (ix86_expand_prologue): Likewise.
526 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
528 PR tree-optimization/82816
529 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
530 if the modes of the two types are the same.
531 (convert_plusminus_to_widen): Likewise.
533 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
535 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
537 (usadv16qi): New define_expand.
538 (usadv8hi): New define_expand.
540 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
543 * ipa-inline-transform.c (update_noncloned_frequencies): Always
545 (inline_transform): Likewise.
546 * predict.c (counts_to_freqs): Remove useless conditional.
547 * profile-count.h (profile_count::apply_scale): Move sanity check.
548 * tree-inline.c (copy_bb): Always scale.
549 (copy_cfg_body): Likewise.
551 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
554 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
556 (*cmp_ite0): Add enabled_for_depr_it attribute.
557 (*cmp_ite1): Likewise.
559 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
561 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
564 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
566 * tree-vrp.c (vrp_int_const_binop): Return true on success and
567 return the value by pointer.
568 (extract_range_from_multiplicative_op_1): Update accordingly.
569 Return as soon as an operation fails.
571 2017-11-05 Tom de Vries <tom@codesourcery.com>
574 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
575 (DEF_SANITIZER_BUILTIN): ... here.
576 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
577 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
579 2017-11-05 Tom de Vries <tom@codesourcery.com>
582 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
584 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
585 ASM_OUTPUT_BEFORE_CASE_LABEL call.
586 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
588 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
589 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
590 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
592 2017-11-05 Tom de Vries <tom@codesourcery.com>
595 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
598 2017-11-04 Michael Clark <michaeljclark@mac.com>
600 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
601 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
603 (*addsi3_extended): Likewise.
604 (*addsi3_extended2): Likewise.
605 (<optab>si3): Likewise.
606 (<optab>di3): Likewise.
607 (<optab><mode>3): Likewise.
608 (<*optabe>si3_internal): Likewise.
609 (zero_extendqi<SUPERQI:mode>2): Likewise.
610 (*add<mode>hi3): Likewise.
611 (*xor<mode>hi3): Likewise.
612 (<optab>di3): Likewise.
613 (*<optab>si3_extend): Likewise.
614 (*sge<u>_<X:mode><GPR:mode>): Likewise.
615 (*slt<u>_<X:mode><GPR:mode>): Likewise.
616 (*sle<u>_<X:mode><GPR:mode>): Likewise.
618 2017-11-04 Andrew Waterman <andrew@sifive.com>
620 * config/riscv/riscv.c (riscv_option_override): Conditionally set
621 TARGET_STRICT_ALIGN based upon -mtune argument.
623 2017-11-04 Andrew Waterman <andrew@sifive.com>
625 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
627 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
629 config/i386/i386.c (choose_basereg): Use optional scratch
630 register and add assertion.
631 (x86_emit_outlined_ms2sysv_save): Use scratch register when
632 needed, and don't allocate stack.
633 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
634 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
635 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
637 2017-11-03 Jeff Law <law@redhat.com>
639 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
640 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
641 to probe at the start of a noreturn function.
643 2017-11-03 Jakub Jelinek <jakub@redhat.com>
645 PR tree-optimization/78821
646 * gimple-ssa-store-merging.c: Update the file comment.
647 (MAX_STORE_ALIAS_CHECKS): Define.
648 (struct store_operand_info): New type.
649 (store_operand_info::store_operand_info): New constructor.
650 (struct store_immediate_info): Add rhs_code and ops data members.
651 (store_immediate_info::store_immediate_info): Add rhscode, op0r
652 and op1r arguments to the ctor, initialize corresponding data members.
653 (struct merged_store_group): Add load_align_base and load_align
655 (merged_store_group::merged_store_group): Initialize them.
656 (merged_store_group::do_merge): Update them.
657 (merged_store_group::apply_stores): Pick the constant for
658 encode_tree_to_bitpos from one of the two operands, or skip
659 encode_tree_to_bitpos if neither operand is a constant.
660 (class pass_store_merging): Add process_store method decl. Remove
661 bool argument from terminate_all_aliasing_chains method decl.
662 (pass_store_merging::terminate_all_aliasing_chains): Remove
663 var_offset_p argument and corresponding handling.
664 (stmts_may_clobber_ref_p): New function.
665 (compatible_load_p): New function.
666 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
667 if there is overlap and rhs_code is not INTEGER_CST. For
668 non-overlapping stores terminate group if rhs is not mergeable.
669 (get_alias_type_for_stmts): Change first argument from
670 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
671 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
672 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
674 (get_location_for_stmts): Change first argument from
675 auto_vec<gimple *> & to vec<gimple *> &.
676 (struct split_store): Remove orig_stmts data member, add orig_stores.
677 (split_store::split_store): Create orig_stores rather than orig_stmts.
678 (find_constituent_stmts): Renamed to ...
679 (find_constituent_stores): ... this. Change second argument from
680 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
681 to info structures rather than the statements.
682 (split_group): Rename ALLOW_UNALIGNED argument to
683 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
684 it. Adjust find_constituent_stores caller.
685 (imm_store_chain_info::output_merged_store): Handle rhs_code other
686 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
687 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
688 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
689 (mem_valid_for_store_merging): New function.
690 (handled_load): New function.
691 (pass_store_merging::process_store): New method.
692 (pass_store_merging::execute): Use process_store method. Adjust
693 terminate_all_aliasing_chains caller.
695 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
697 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
698 Return true for more constants, symbols and label references.
699 (aarch64_valid_floating_const): Remove unused function.
701 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
704 * config/aarch64/aarch64.c (aarch64_layout_frame):
705 Undo forcing of LR at bottom of frame.
707 2017-11-03 Jeff Law <law@redhat.com>
710 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
711 for int_registers_saved.
713 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
714 extracted from tree-ssa-dom.c.
715 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
716 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
717 (record_equivalences_from_incoming_edge): Add additional argument
718 to single_pred_edge_ignoring_loop_edges call.
719 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
720 (uncprop_dom_walker::before_dom_children): Add additional argument
721 to single_pred_edge_ignoring_loop_edges call.
722 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
723 single_pred_edge_ignoring_loop_edges rather than open coding.
724 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
726 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
728 * match.pd (-(-A)): Rewrite.
730 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
732 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
733 (rs6000_emit_int_cmove): New declaration.
734 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
735 (rs6000_emit_sISEL): Delete.
736 (rs6000_emit_int_cmove): Make non-static.
737 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
738 instead of rs6000_emit_sISEL.
740 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
742 * asan.c (create_cond_insert_point): Maintain profile.
743 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
745 * basic-block.h (struct basic_block_def): Remove frequency.
746 (EDGE_FREQUENCY): Use to_frequency
747 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
749 (find_traces): Update to use to_frequency.
750 (find_traces_1_round): Likewise; use only IPA counts.
751 (bb_to_key): Likewise.
752 (connect_traces): Use IPA counts only.
753 (copy_bb_p): Update to use to_frequency.
754 (fix_up_crossing_landing_pad): Likewise.
755 (sanitize_hot_paths): Likewise.
756 * bt-load.c (basic_block_freq): Likewise.
757 * cfg.c (init_flow): Set count_max to uninitialized.
758 (check_bb_profile): Remove frequencies; check counts.
759 (dump_bb_info): Do not dump frequencies.
760 (update_bb_profile_for_threading): Update counts only.
761 (scale_bbs_frequencies_int): Likewise.
762 (MAX_SAFE_MULTIPLIER): Remove.
763 (scale_bbs_frequencies_gcov_type): Update counts only.
764 (scale_bbs_frequencies_profile_count): Update counts only.
765 (scale_bbs_frequencies): Update counts only.
766 * cfg.h (struct control_flow_graph): Add count-max.
767 (update_bb_profile_for_threading): Update prototype.
768 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
769 (find_many_sub_basic_blocks): Likewise.
770 * cfgcleanup.c (try_forward_edges): Likewise.
771 (try_crossjump_to_edge): Likewise.
772 * cfgexpand.c (expand_gimple_cond): Likewise.
773 (expand_gimple_tailcall): Likewise.
774 (construct_init_block): Likewise.
775 (construct_exit_block): Likewise.
776 * cfghooks.c (verify_flow_info): Check consistency of counts.
777 (dump_bb_for_graph): Do not dump frequencies.
778 (split_block_1): Do not update frequencies.
779 (split_edge): Do not update frequencies.
780 (make_forwarder_block): Do not update frequencies.
781 (duplicate_block): Do not update frequencies.
782 (account_profile_record): Do not update frequencies.
783 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
784 for global heuristics.
785 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
786 (expected_loop_iterations_unbounded): Use counts only.
787 * cfgloopmanip.c (scale_loop_profile): Simplify.
788 (create_empty_loop_on_edge): Simplify
790 (duplicate_loop_to_header_edge): Simplify
791 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
792 (update_br_prob_note): Take care of removing note when profile
794 (relink_block_chain): Do not dump frequency.
795 (rtl_account_profile_record): Use to_frequency.
796 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
797 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
798 (cgraph_update_edges_for_call_stmt_node): Likewise.
799 (cgraph_edge::verify_count_and_frequency): Update.
800 (cgraph_node::verify_node): Temporarily disable frequency verification.
801 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
803 (cgraph_edge::rebuild_edges): Convert to ipa counts.
804 * cgraphunit.c (init_lowered_empty_function): Do not initialize
806 (cgraph_node::expand_thunk): Update profile.
807 * except.c (dw2_build_landing_pads): Do not update frequency.
808 * final.c (compute_alignments): Use to_frequency.
809 (dump_basic_block_info): Do not dump frequency.
810 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
811 (dump_gimple_bb_header): Do not dump frequency.
812 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
814 * gimple-streamer-in.c (input_bb): Do not stream frequency.
815 * gimple-streamer-out.c (output_bb): Do not stream frequency.
816 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
817 (init_before_recovery): Do not update frequency.
818 (sched_create_recovery_edges): Do not update frequency.
819 * hsa-gen.c (convert_switch_statements): Do not update frequency.
820 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
821 (ipa_cp_c_finalize): Set max_count to uninitialized.
822 * ipa-fnsummary.c (get_minimal_bb): Use counts.
823 (param_change_prob): Use counts.
824 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
826 * ipa-split.c (consider_split): Use to_frequency.
827 (split_function): Use to_frequency.
828 * ira-build.c (loop_compare_func): Likewise.
829 (mark_loops_for_removal): Likewise.
830 (mark_all_loops_for_removal): Likewise.
831 * loop-doloop.c (doloop_modify): Do not update frequency.
832 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
834 * lto-streamer-in.c (input_function): Update count_max.
835 * omp-expand.c (expand_omp_taskreg): Update count_max.
836 * omp-simd-clone.c (simd_clone_adjust): Update profile.
837 * predict.c (maybe_hot_frequency_p): Use to_frequency.
838 (maybe_hot_count_p): Use ipa counts only.
839 (maybe_hot_bb_p): Simplify.
840 (maybe_hot_edge_p): Simplify.
841 (probably_never_executed): Do not take frequency argument.
842 (probably_never_executed_bb_p): Do not pass frequency.
843 (probably_never_executed_edge_p): Likewise.
844 (combine_predictions_for_bb): Check that profile is nonzero.
845 (propagate_freq): Do not set frequency.
846 (drop_profile): Simplify.
847 (counts_to_freqs): Simplify.
848 (expensive_function_p): Use to_frequency.
849 (propagate_unlikely_bbs_forward): Simplify.
850 (determine_unlikely_bbs): Simplify.
851 (estimate_bb_frequencies): Add hack to silence graphite issues.
852 (compute_function_frequency): Use ipa counts.
853 (pass_profile::execute): Update.
854 (rebuild_frequencies): Use counts only.
855 (force_edge_cold): Use counts only.
856 * profile-count.c (profile_count::dump): Dump new count types.
857 (profile_count::differs_from_p): Check compatiblity.
858 (profile_count::to_frequency): New function.
859 (profile_count::to_cgraph_frequency): New function.
860 * profile-count.h (struct function): Declare.
861 (enum profile_quality): Add profile_guessed_local and
862 profile_guessed_global0.
863 (class profile_proability): Decrease number of bits to 29;
864 update from_reg_br_prob_note and to_reg_br_prob_note.
865 (class profile_count: Update comment; decrease number of bits
866 to 61. Check compatibility.
867 (profile_count::compatible_p): New private member function.
868 (profile_count::ipa_p): New member function.
869 (profile_count::operator<): Handle global zero correctly.
870 (profile_count::operator>): Handle global zero correctly.
871 (profile_count::operator<=): Handle global zero correctly.
872 (profile_count::operator>=): Handle global zero correctly.
873 (profile_count::nonzero_p): New member function.
874 (profile_count::force_nonzero): New member function.
875 (profile_count::max): New member function.
876 (profile_count::apply_scale): Handle IPA scalling.
877 (profile_count::guessed_local): New member function.
878 (profile_count::global0): New member function.
879 (profile_count::ipa): New member function.
880 (profile_count::to_frequency): Declare.
881 (profile_count::to_cgraph_frequency): Declare.
882 * profile.c (OVERLAP_BASE): Delete.
883 (compute_frequency_overlap): Delete.
884 (compute_branch_probabilities): Do not use compute_frequency_overlap.
885 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
886 * sched-ebb.c (rank): Use counts only.
887 * shrink-wrap.c (handle_simple_exit): Use counts only.
888 (try_shrink_wrapping): Use counts only.
889 (place_prologue_for_one_component): Use counts only.
890 * tracer.c (find_best_predecessor): Use to_frequency.
891 (find_trace): Use to_frequency.
892 (tail_duplicate): Use to_frequency.
893 * trans-mem.c (expand_transaction): Do not update frequency.
894 * tree-call-cdce.c: Do not update frequency.
895 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
896 (gimple_merge_blocks): Likewise.
897 (gimple_split_edge): Likewise.
898 (gimple_duplicate_sese_region): Likewise.
899 (gimple_duplicate_sese_tail): Likewise.
900 (move_sese_region_to_fn): Likewise.
901 (gimple_account_profile_record): Likewise.
902 (insert_cond_bb): Likewise.
903 * tree-complex.c (expand_complex_div_wide): Likewise.
904 * tree-eh.c (lower_resx): Update profile.
905 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
907 (initialize_cfun): Do not initialize frequencies
908 (freqs_to_counts): Delete.
909 (copy_cfg_body): Ignore count parameter.
911 (expand_call_inline): Update count_max.
912 (optimize_inline_calls): Update count_max.
913 (tree_function_versioning): Update count_max.
914 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
915 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
917 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
918 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
919 (try_peel_loop): Likewise.
920 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
922 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
923 (tree_transform_and_unroll_loop): Do not use frequencies
924 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
925 Use reliable prediction only.
926 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
927 * tree-ssa-sink.c (select_best_block): Use to_frequency.
928 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
930 * tree-ssa-threadupdate.c (create_block_for_threading): Do
932 (any_remaining_duplicated_blocks): Likewise.
933 (update_profile): Likewise.
934 (estimated_freqs_path): Delete.
935 (freqs_to_counts_path): Delete.
936 (clear_counts_path): Delete.
937 (ssa_fix_duplicate_block_edges): Likewise.
938 (duplicate_thread_path): Likewise.
939 * tree-switch-conversion.c (gen_inbound_check): Use counts.
940 * tree-tailcall.c (decrease_profile): Do not update frequency.
941 (eliminate_tail_call): Likewise.
942 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
943 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
944 (optimize_mask_stores): Likewise.
945 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
946 * ubsan.c (ubsan_expand_null_ifn): Update profile.
947 (ubsan_expand_ptr_ifn): Update profile.
948 * value-prof.c (gimple_ic): Simplify.
949 * value-prof.h (gimple_ic): Update prototype.
950 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
951 * ipa-inline.c (compute_uninlined_call_time): Be sure that
953 (want_inline_self_recursive_call_p): Likewise.
954 (resolve_noninline_speculation): Only cummulate defined counts.
955 (inline_small_functions): Use nonzero_p.
956 (ipa_inline): Do not access freed node.
958 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
960 * config/aarch64/aarch64.c (aarch64_override_options_internal):
961 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
963 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
965 * config/riscv/riscv.c (riscv_legitimize_move): Handle
966 non-legitimate address.
968 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
970 * config/rs6000/rs6000.md (*lt0_disi): Delete.
971 (*lt0_<mode>di, *lt0_<mode>si): New.
973 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
975 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
978 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
979 Jim Wilson <jim.wilson@linaro.org>
981 * config/aarch64/aarch64-cores.def (saphira): New CPU.
982 * config/aarch64/aarch64-tune.md: Regenerated.
983 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
984 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
986 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
988 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
989 (arc_expand_prologue): Restore blink for millicode.
990 * config/arc/linux.h (LINK_EH_SPEC): Defined.
992 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
995 * config/i386/i386.c (ix86_vector_duplicate_value): Use
996 gen_vec_duplicate after forcing the scalar into a register.
998 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1000 * combine (try_combine): Print the insns input to try_combine to the
1003 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1006 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1007 Remove second argument from aarch64_process_target_attr call.
1008 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1010 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1012 (aarch64_handle_attr_arch): Remove second argument.
1013 (aarch64_handle_attr_cpu): Ditto.
1014 (aarch64_handle_attr_tune): Ditto.
1015 (aarch64_handle_attr_isa_flags): Ditto.
1016 (aarch64_process_one_target_attr): Ditto.
1017 (aarch64_process_target_attr): Ditto.
1018 (aarch64_option_valid_attribute_p): Remove second argument.
1019 on aarch64_process_target_attr call.
1021 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1023 * diagnostic.c: Include "selftest-diagnostic.h".
1024 (selftest::assert_location_text): New function.
1025 (selftest::test_diagnostic_get_location_text): New function.
1026 (selftest::diagnostic_c_tests): Call it.
1028 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1030 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1031 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1032 (class selftest::test_diagnostic_context): Move to...
1033 * selftest-diagnostic.c: New file.
1034 * selftest-diagnostic.h: New file.
1036 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1038 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1039 offset range for FT32B.
1040 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1041 * config/ft32/ft32.md: Add TARGET_NOPM.
1042 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1043 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1045 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1047 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1049 2017-11-02 Jeff Law <law@redhat.com>
1051 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1052 virtual keyword on FINAL OVERRIDE members.
1054 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1055 virtuals together. Add virtual destructor.
1056 (substitute_and_fold_engine): Similarly.
1058 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1060 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1062 2017-11-02 Richard Biener <rguenther@suse.de>
1064 PR tree-optimization/82795
1065 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1067 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1069 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1071 (gcc_GAS_CHECK_FEATURE): Remove.
1072 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1073 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1074 * configure: Regenerate.
1076 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1078 * config/arc/arc.c (hwloop_optimize): Account for empty
1081 2017-11-02 Richard Biener <rguenther@suse.de>
1084 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1085 Truncate ARRAY_REF index and element size.
1087 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1089 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1091 2017-11-01 Jeff Law <law@redhat.com>
1093 * tree-ssa-ccp.c (ccp_folder): New class derived from
1094 substitute_and_fold_engine.
1095 (ccp_folder::get_value): New member function.
1096 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1097 (ccp_fold_stmt): Remove prototype.
1098 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1099 * tree-ssa-copy.c (copy_folder): New class derived from
1100 substitute_and_fold_engine.
1101 (copy_folder::get_value): Renamed from get_value.
1102 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1103 * tree-vrp.c (vrp_folder): New class derived from
1104 substitute_and_fold_engine.
1105 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1106 (vrp_folder::get_value): New member function.
1107 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1108 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1109 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1110 provide a class interface to folder/substitute routines.
1111 (ssa_prop_fold_stmt_fn): Remove typedef.
1112 (ssa_prop_get_value_fn): Likewise.
1113 (subsitute_and_fold): Remove prototype.
1114 (replace_uses_in): Likewise.
1115 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1116 Renamed from replace_uses_in. Call the virtual member function
1117 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1118 (substitute_and_fold_dom_walker): Remove initialization of
1119 data member entries for calbacks. Add substitute_and_fold_engine
1120 member and initialize it.
1121 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1122 member functions for get_value, replace_phi_args_in c
1123 replace_uses_in, and fold_stmt calls.
1124 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1125 substitute_and_fold. Remove assert. Update ctor call.
1127 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1128 (ssa_prop_visit_phi_fn): Likewise.
1129 (class ssa_propagation_engine): New class to provide an interface
1131 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1133 (ssa_prop_visit_phi): Likewise.
1134 (ssa_propagation_engine::simulate_stmt): Moved into class.
1135 Call visit_phi/visit_stmt from the class rather than via
1136 file scoped static variables.
1137 (ssa_propagation_engine::simulate_block): Moved into class.
1138 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1139 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1140 set file scoped statics for the visit_stmt/visit_phi callbacks.
1141 * tree-complex.c (complex_propagate): New class derived from
1142 ssa_propagation_engine.
1143 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1144 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1145 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1147 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1148 ssa_propagation_engine.
1149 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1150 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1151 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1152 * tree-ssa-copy.c (copy_prop): New class derived from
1153 ssa_propagation_engine.
1154 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1155 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1156 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1157 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1158 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1159 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1160 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1162 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1164 PR rtl-optimization/82778
1165 PR rtl-optimization/82597
1166 * compare-elim.c (struct comparison): Add in_a_setter field.
1167 (find_comparison_dom_walker::before_dom_children): Remove killed
1168 bitmap and df_simulate_find_defs call, instead walk the defs.
1169 Compute last_setter and initialize in_a_setter. Merge definitions
1170 with first initialization for a few variables.
1171 (try_validate_parallel): Use insn_invalid_p instead of
1172 recog_memoized. Return insn rather than just the pattern.
1173 (try_merge_compare): Fix up comment. Don't uselessly test if
1174 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1176 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1179 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1180 Alan Hayward <alan.hayward@arm.com>
1181 David Sherwood <david.sherwood@arm.com>
1183 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1184 aarch64_hard_regno_nregs to get the number of registers
1187 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1188 Alan Hayward <alan.hayward@arm.com>
1189 David Sherwood <david.sherwood@arm.com>
1191 * config/aarch64/constraints.md (Upl): Rename to...
1193 * config/aarch64/aarch64.md
1194 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1197 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1198 Alan Hayward <alan.hayward@arm.com>
1199 David Sherwood <david.sherwood@arm.com>
1201 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1202 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1205 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1206 Alan Hayward <alan.hayward@arm.com>
1207 David Sherwood <david.sherwood@arm.com>
1209 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1210 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1211 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1213 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1214 of a single element.
1215 * config/aarch64/iterators.md: Add a comment above the permute
1216 unspecs to say that they are generated directly by
1217 aarch64_expand_vec_perm_const.
1218 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1220 2017-11-01 Nathan Sidwell <nathan@acm.org>
1222 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1224 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1226 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1227 and medany code models, and describe what they do.
1229 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1231 Revert accidental duplicate:
1233 * combine.c (can_change_dest_mode): Reject changes in
1234 REGMODE_NATURAL_SIZE.
1236 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1238 PR rtl-optimization/64682
1239 PR rtl-optimization/69567
1240 PR rtl-optimization/69737
1241 PR rtl-optimization/82683
1242 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1243 register mentioned in the note, drop the note, unless it came from I3,
1244 in which case it should go to I3 again.
1246 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1248 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1249 and return false if not.
1250 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1252 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1254 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1255 range is known to be empty.
1257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1258 Alan Hayward <alan.hayward@arm.com>
1259 David Sherwood <david.sherwood@arm.com>
1261 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1262 and CONST_VECTOR_NUNITS instead of computing the number of units from
1263 the byte sizes of the vector and element.
1264 (simplify_binary_operation_1): Likewise.
1265 (simplify_const_binary_operation): Likewise.
1266 (simplify_ternary_operation): Likewise.
1268 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1273 (int_mem_offset): ...this new function.
1274 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1275 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1276 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1279 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1280 Alan Hayward <alan.hayward@arm.com>
1281 David Sherwood <david.sherwood@arm.com>
1283 * lower-subreg.c (interesting_mode_p): New function.
1284 (compute_costs, find_decomposable_subregs, decompose_register)
1285 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1286 (resolve_clobber, dump_choices): Use it.
1288 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1289 Alan Hayward <alan.hayward@arm.com>
1290 David Sherwood <david.sherwood@arm.com>
1292 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1294 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1295 Alan Hayward <alan.hayward@arm.com>
1296 David Sherwood <david.sherwood@arm.com>
1298 * alias.c (find_base_value, find_base_term): Only process integer
1299 truncations. Check the precision rather than the size.
1301 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1302 Alan Hayward <alan.hayward@arm.com>
1303 David Sherwood <david.sherwood@arm.com>
1305 * machmode.h (is_narrower_int_mode): New function
1306 * optabs.c (expand_float, expand_fix): Use it.
1307 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1309 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1310 Alan Hayward <alan.hayward@arm.com>
1311 David Sherwood <david.sherwood@arm.com>
1313 * rtl.h (narrower_subreg_mode): New function.
1314 * ira-color.c (update_costs_from_allocno): Use it.
1316 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1317 Alan Hayward <alan.hayward@arm.com>
1318 David Sherwood <david.sherwood@arm.com>
1320 * optabs-query.h (convert_optab_p): New function, split out from...
1321 (convert_optab_handler): ...here.
1322 (widening_optab_handler): Delete.
1323 (find_widening_optab_handler): Remove permit_non_widening parameter.
1324 (find_widening_optab_handler_and_mode): Likewise. Provide an
1325 override that operates on mode class wrappers.
1326 * optabs-query.c (widening_optab_handler): Delete.
1327 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1328 parameter. Assert that the two modes are the same class and that
1329 the "from" mode is narrower than the "to" mode. Use
1330 convert_optab_handler instead of widening_optab_handler.
1331 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1332 instead of widening_optab_handler.
1333 * expr.c (expand_expr_real_2): Update calls to
1334 find_widening_optab_handler.
1335 * optabs.c (expand_widen_pattern_expr): Likewise.
1336 (expand_binop_directly): Take the insn_code as a parameter.
1337 (expand_binop): Only call find_widening_optab_handler for
1338 conversion optabs; use optab_handler otherwise. Update calls
1339 to find_widening_optab_handler and expand_binop_directly.
1340 Use convert_optab_handler instead of widening_optab_handler.
1341 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1342 find_widening_optab_handler and use scalar_mode rather than
1344 (convert_plusminus_to_widen): Likewise.
1346 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1347 Alan Hayward <alan.hayward@arm.com>
1348 David Sherwood <david.sherwood@arm.com>
1350 * machmode.h (fixed_size_mode): New class.
1351 * rtl.h (get_pool_mode): Return fixed_size_mode.
1352 * gengtype.c (main): Add fixed_size_mode.
1353 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1354 (get_raw_arg_mode): Likewise.
1355 * doc/tm.texi: Regenerate.
1356 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1357 * targhooks.c (default_get_reg_raw_mode): Likewise.
1358 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1359 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1360 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1361 (msp430_get_raw_result_mode): Likewise.
1362 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1363 * dbxout.c (dbxout_parms): Require fixed-size modes.
1364 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1365 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1366 * omp-low.c (lower_oacc_reductions): Likewise.
1367 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1368 (simplify_subreg): Update accordingly.
1369 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1370 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1371 that aren't fixed-size.
1372 (get_pool_mode): Return a fixed_size_mode.
1373 (output_constant_pool_2): Take a fixed_size_mode.
1375 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1376 Alan Hayward <alan.hayward@arm.com>
1377 David Sherwood <david.sherwood@arm.com>
1379 * doc/rtl.texi (vec_series): Document.
1380 (const): Say that the operand can be a vec_series.
1381 * rtl.def (VEC_SERIES): New rtx code.
1382 * rtl.h (const_vec_series_p_1): Declare.
1383 (const_vec_series_p): New function.
1384 * emit-rtl.h (gen_const_vec_series): Declare.
1385 (gen_vec_series): Likewise.
1386 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1387 (gen_vec_series): Likewise.
1388 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1389 * simplify-rtx.c (simplify_unary_operation): Handle negations
1391 (simplify_binary_operation_series): New function.
1392 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1393 (test_vector_ops_series): New function.
1394 (test_vector_ops): Call it.
1395 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1396 gen_const_vec_series.
1397 (altivec_lvsr): Likewise.
1398 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1400 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1401 Alan Hayward <alan.hayward@arm.com>
1402 David Sherwood <david.sherwood@arm.com>
1404 * doc/rtl.texi (const): Update description of address constants.
1405 Say that vector constants are allowed too.
1406 * common.md (E, F): Use CONSTANT_P instead of checking for
1408 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1409 checking for CONST_VECTOR.
1410 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1412 * expr.c (expand_expr_real_2): Check for vector modes instead
1413 of checking for CONST_VECTOR.
1414 * rtl.h (const_vec_p): New function.
1415 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1416 (unwrap_const_vec_duplicate): Handle them here too.
1418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1419 David Malcolm <dmalcolm@redhat.com>
1420 Alan Hayward <alan.hayward@arm.com>
1421 David Sherwood <david.sherwood@arm.com>
1423 * rtl.h (vec_duplicate_p): New function.
1424 * selftest-rtl.c (assert_rtx_eq_at): New function.
1425 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1426 (assert_rtx_eq_at): Declare.
1427 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1428 * selftest-run-tests.c (selftest::run_tests): Call it.
1429 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1430 (simplify_unary_operation_1): Recursively handle vector duplicates.
1431 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1433 (simplify_subreg): Handle subregs of vector duplicates.
1434 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1435 (selftest::simplify_rtx_c_tests): New functions.
1437 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1438 Alan Hayward <alan.hayward@arm.com>
1439 David Sherwood <david.sherwood@arm.com>
1441 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1442 (gen_vec_duplicate): Likewise.
1443 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1445 (gen_const_vector): ...here.
1446 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1447 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1448 whose elements are all equal.
1449 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1450 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1451 (simplify_relational_operation): Likewise.
1452 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1454 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1455 (aarch64_expand_vector_init): Likewise.
1456 * config/arm/arm.c (neon_vdup_constant): Likewise.
1457 (neon_expand_vector_init): Likewise.
1458 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1459 (arm_block_set_unaligned_vect): Likewise.
1460 (arm_block_set_aligned_vect): Likewise.
1461 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1462 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1463 (expand_vec_perm_even_odd_pack): Likewise.
1464 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1465 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1466 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1467 gen_const_vec_duplicate.
1468 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1469 * config/mips/mips.c (mips_gen_const_int_vector): Use
1470 gen_const_vec_duplicate.
1471 (mips_expand_vector_init): Use CONST0_RTX.
1472 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1473 (define_split): Use gen_const_vec_duplicate.
1474 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1475 (define_split): Use gen_const_vec_duplicate.
1476 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1477 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1478 * config/spu/spu.c (spu_const): Likewise.
1480 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * combine.c (can_change_dest_mode): Reject changes in
1485 REGMODE_NATURAL_SIZE.
1487 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1489 * configure.ac (--enable-libssp): New.
1490 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1491 trying to determine target-specific default. Adjust indentation.
1492 * configure: Regenerated.
1493 * doc/install.texi (Configuration): Expand --disable-libssp
1496 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1498 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1501 2017-10-31 Martin Jambor <mjambor@suse.cz>
1504 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1506 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1508 * auto-profile.c (autofdo_source_profile::read): Use
1509 UNKNOWN_LOCATION rather than 0.
1510 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1511 (warning_at): ...this overload.
1512 (warning_at_rich_loc_n): Rename to...
1513 (warning_n): ...this overload.
1514 (error_at_rich_loc): Rename to...
1515 (error_at): ...this overload.
1516 (pedwarn_at_rich_loc): Rename to...
1517 (pedwarn): ...this overload.
1518 (permerror_at_rich_loc): Rename to...
1519 (permerror): ...this overload.
1520 (inform_at_rich_loc): Rename to...
1521 (inform): ...this overload.
1522 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1523 (diagnostic_n_impl_richloc): Rename to...
1524 (diagnostic_n_impl): ...this rich_location *-based decl.
1525 (inform_at_rich_loc): Rename to...
1526 (inform): ...this, and add an assertion.
1527 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1528 (warning_at_rich_loc): Rename to...
1529 (warning_at): ...this, and add an assertion.
1530 (warning_at_rich_loc_n): Rename to...
1531 (warning_n): ...this, and add an assertion.
1532 (warning_n): Update location_t-based implementation for removal of
1533 location_t-based diagnostic_n_impl.
1534 (pedwarn_at_rich_loc): Rename to...
1535 (pedwarn): ...this, and add an assertion.
1536 (permerror_at_rich_loc): Rename to...
1537 (permerror): ...this, and add an assertion.
1538 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1539 (error_at_rich_loc): Rename to...
1540 (error_at): ...this, and add an assertion.
1541 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1542 (driver::do_spec_on_infiles): Likewise.
1543 * substring-locations.c (format_warning_va): Update for renaming
1544 of inform_at_rich_loc.
1546 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1548 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1549 _Float<N>X built-in functions so that the variant without the
1550 "__builtin_" prefix is only enabled for the GNU C and Objective C
1551 languages when they are in non-strict ANSI/ISO mode.
1552 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1553 * target.def (floatn_builtin_p): Add a target hook to control
1554 whether _Float<N> and _Float<N>X built-in functions without the
1555 "__builtin_" prefix are enabled, and return true for C and
1556 Objective C in the default hook. Include langhooks.h in
1558 * targhooks.h (default_floatn_builtin_p): Likewise.
1559 * targhooks.c (default_floatn_builtin_p): Likewise.
1560 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1561 floatn_builtin_p target hook.
1562 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1564 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1565 Eric Botcazou <ebotcazou@adacore.com>
1567 PR rtl-optimization/81803
1568 * lra-constraints.c (curr_insn_transform): Also reload the whole
1569 register for a strict subreg no wider than a word if this is for
1570 a WORD_REGISTER_OPERATIONS target.
1572 2017-10-31 Jason Merrill <jason@redhat.com>
1574 * gdbinit.in: Skip over inlines from timevar.h.
1576 2017-10-31 Martin Liska <mliska@suse.cz>
1578 * doc/gcov.texi: Document new option.
1579 * gcov.c (print_usage): Likewise print it.
1580 (process_args): Support the argument.
1581 (format_count): New function.
1582 (format_gcov): Use the function.
1584 2017-10-31 Martin Liska <mliska@suse.cz>
1586 * gcov.c (struct name_map): do not use typedef.
1587 Define operator== and operator<.
1588 (name_search): Remove.
1589 (name_sort): Remove.
1590 (main): Do not allocate names.
1591 (process_file): Add vertical space.
1592 (generate_results): Use std::find.
1593 (release_structures): Do not release memory.
1594 (find_source): Use std::find.
1596 2017-10-31 Martin Liska <mliska@suse.cz>
1598 * gcov.c (struct line_info): Remove it's typedef.
1599 (line_info::line_info): Add proper ctor.
1600 (line_info::has_block): Do not use a typedef.
1601 (struct source_info): Do not use typedef.
1602 (circuit): Likewise.
1603 (get_cycles_count): Likewise.
1604 (output_intermediate_file): Iterate via vector iterator.
1605 (add_line_counts): Use std::vector methods.
1606 (accumulate_line_counts): Likewise.
1607 (output_lines): Likewise.
1609 2017-10-31 Martin Liska <mliska@suse.cz>
1611 * gcov.c (struct source_info): Remove typedef.
1612 (source_info::source_info): Add proper ctor.
1613 (accumulate_line_counts): Use struct, not it's typedef.
1614 (output_gcov_file): Likewise.
1615 (output_lines): Likewise.
1616 (main): Do not allocate an array.
1617 (output_intermediate_file): Use size of vector container.
1618 (process_file): Resize the vector.
1619 (generate_results): Do not preallocate, use newly added vector
1621 (release_structures): Do not release sources.
1622 (find_source): Use vector methods.
1623 (add_line_counts): Do not use typedef.
1625 2017-10-31 Martin Liska <mliska@suse.cz>
1627 * doc/gcov.texi: Document that.
1628 * gcov.c (add_line_counts): Mark lines with a non-executed
1630 (output_line_beginning): Handle such lines.
1631 (output_lines): Pass new argument.
1632 (output_intermediate_file): Print it in intermediate format.
1634 2017-10-31 Martin Liska <mliska@suse.cz>
1636 * color-macros.h: New file.
1637 * diagnostic-color.c: Factor out color related to macros to
1639 * doc/gcov.texi: Document -k option.
1640 * gcov.c (INCLUDE_STRING): Include string.h.
1641 (print_usage): Add -k option.
1642 (process_args): Parse it.
1643 (pad_count_string): New function.
1644 (output_line_beginning): Likewise.
1645 (DEFAULT_LINE_START): New macro.
1646 (output_lines): Support color output.
1648 2017-10-31 Martin Liska <mliska@suse.cz>
1650 PR gcov-profile/82633
1651 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1652 their interaction with GCOV infrastructure.
1653 * configure.ac: Add -fkeep-{inline,static}-functions to
1655 * configure: Regenerate.
1657 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1660 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1662 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1665 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1666 into a register if it does not fit into an immediate offset field.
1668 2017-10-31 Olivier Hainque <hainque@adacore.com>
1670 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1672 2017-10-31 Julia Koval <julia.koval@intel.com>
1674 * config.gcc: Add gfniintrin.h.
1675 * config/i386/gfniintrin.h: New.
1676 * config/i386/i386-builtin-types.def
1677 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1678 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1679 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1680 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1681 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1682 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1683 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1684 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1685 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1686 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1687 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1688 * config/i386/immintrin.h: Include gfniintrin.h.
1689 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1691 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1693 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1695 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1697 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1698 (arm_ashldi3_1bit): Remove pattern.
1699 (ashrdi3): Remove shift by 1 expansion.
1700 (arm_ashrdi3_1bit): Remove pattern.
1701 (lshrdi3): Remove shift by 1 expansion.
1702 (arm_lshrdi3_1bit): Remove pattern.
1703 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1704 cost of ashldi3 by 1.
1705 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1706 (<shift>di3_neon): Likewise.
1708 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1710 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1711 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1712 and (*aarch64_simd_mov<VQ:mode>).
1713 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1714 pattern alternative.
1715 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1716 attributes to match pattern alternative.
1718 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1720 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1721 * config/rs6000/emmintrin.h: New file.
1722 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1724 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1726 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1727 * (movdi_vfp_cortexa8): Remove pattern.
1729 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1731 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1733 (Specific, alpha*-dec-osf5.1): Remove.
1734 (Specific, mips-sgi-irix5): Remove.
1735 (Specific, mips-sgi-irix6): Remove.
1737 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1740 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1741 arguments to clear_bit_region_be.
1743 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1745 * gimplify.c: Include memmodel.h.
1747 2017-10-30 Martin Jambor <mjambor@suse.cz>
1749 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1750 condition whether loop should be executed at all.
1752 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1755 gimple folding of vec_madd() intrinsics.
1756 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1757 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1758 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1760 2017-10-30 Richard Biener <rguenther@suse.de>
1762 PR tree-optimization/82762
1764 2017-10-23 Richard Biener <rguenther@suse.de>
1766 PR tree-optimization/82129
1768 2017-08-01 Richard Biener <rguenther@suse.de>
1770 PR tree-optimization/81181
1771 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1772 (compute_antic): ... end of iteration here.
1774 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1776 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1777 -std=iso9899:2017 and -std=gnu17.
1778 * doc/standards.texi (C Language): Document C17 support.
1779 * doc/cpp.texi (Overview): Mention -std=c17.
1780 (Standard Predefined Macros): Document C11 and C17 values of
1781 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1782 * doc/extend.texi (Inline): Do not list individual options for
1783 standards newer than C99.
1784 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1786 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1789 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1791 * asan.c (asan_finish_file): Align asan globals array by shadow
1794 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1797 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1798 (struct store_immediate_info): Add bitregion_start and bitregion_end
1800 (store_immediate_info::store_immediate_info): Add brs and bre
1801 arguments and initialize bitregion_{start,end} from those.
1802 (struct merged_store_group): Add bitregion_start, bitregion_end,
1803 align_base and mask fields. Drop unnecessary struct keyword from
1804 struct store_immediate_info. Add do_merge method.
1805 (clear_bit_region_be): Use memset instead of loop storing zeros.
1806 (merged_store_group::do_merge): New method.
1807 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1808 stores as long as the surrounding bitregions have no gaps.
1809 (merged_store_group::merge_overlapping): Use do_merge.
1810 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1811 is byte aligned, rather than requiring that start and width are
1812 byte aligned. Drop unnecessary struct keyword from
1813 struct store_immediate_info. Allocate and populate also mask array.
1814 Make start of the arrays relative to bitregion_start rather than
1815 start and size them according to bitregion_{end,start} difference.
1816 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1817 struct store_immediate_info.
1818 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1819 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1820 struct keyword from struct store_immediate_info.
1821 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1822 between stores as long as the surrounding bitregions have no gaps.
1824 (struct split_store): Add orig non-static data member.
1825 (split_store::split_store): Initialize orig to false.
1826 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1827 if there is exactly a single original stmt. Change stmts argument
1828 to pointer from reference, if NULL, don't push anything to it. Add
1829 first argument, use it to optimize skipping over orig stmts that
1830 are known to be before bitpos already. Simplify.
1831 (split_group): Return unsigned int count how many stores are or
1832 would be needed rather than a bool. Add allow_unaligned argument.
1833 Change split_stores argument from reference to pointer, if NULL,
1834 only do a dry run computing how many stores would be produced.
1835 Rewritten algorithm to use both alignment and misalign if
1836 !allow_unaligned and handle bitfield stores with gaps.
1837 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1838 from bitregion_start instead of start. Compute allow_unaligned
1839 here, if true, do 2 split_group dry runs to compute which one
1840 produces fewer stores and prefer aligned if equal. Punt if
1841 new count is bigger or equal than original before emitting any
1842 statements, rather than during that. Remove no longer needed
1843 new_ssa_names tracking. Replace num_stmts with
1844 split_stores.length (). Use 32-bit stack allocated entries
1845 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1846 if possible. Handle bitfields with gaps.
1847 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1848 Compute bitregion_{start,end} for the stores and construct
1849 store_immediate_info with that. Formatting fixes.
1851 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1854 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1855 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1857 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1859 * gimplify.c: Include tm_p.h.
1861 * common.opt (gcoff): Re-add as ignored option.
1862 (gcoff1, gcoff2, gcoff3): Likewise.
1864 * Makefile.in (OBJS): Delete sdbout.o.
1865 (GTFILES): Delete $(srcdir)/sdbout.c.
1866 * debug.h: Delete sdb_debug_hooks.
1867 * final.c: Delete sdbout.h include.
1868 (final_scan_insn): Delete SDB_DEBUG check.
1869 (rest_of_clean_state): Likewise.
1870 * output.h: Delete sdb_begin_function_line.
1873 * toplev.c: Delete sdbout.h include.
1874 (process_options): Delete SDB_DEBUG check.
1875 * tree-core.h (tree_type_common): Delete pointer field of
1877 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1878 TYPE_SYMTAB_POINTER.
1879 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1880 (TYPE_SYMTAB_IS_POINTER): Delete.
1881 (TYPE_SYMTAB_IS_DIE): Renumber.
1882 * xcoffout.c: Refer to former sdbout.c file.
1883 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1885 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1886 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1887 (Debugging Options): Delete -gcoff.
1888 (-gstabs): Delete SDB reference.
1890 (-gcoff@var{level}): Delete.
1891 * doc/passes.texi (Debugging information output): Delete SDB and
1892 sdbout.c references.
1893 * doc/tm.texi: Regenerate.
1894 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1895 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1897 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1898 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1899 (SDB_DEBUGGING_INFO): Delete.
1900 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1901 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1902 * target.def (output_source_filename): Delete COFF reference.
1904 * common.opt (gcoff): Delete.
1905 (gxcoff+): Update Negative chain.
1906 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1908 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1909 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1910 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1912 (expand_function_start): Change sdb reference to past tense.
1913 (expand_function_end): Change sdb reference to past tense.
1914 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1915 * opts.c (debug_type_names): Delete coff entry.
1916 (common_handle_option): Delete OPT_gcoff case.
1917 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1919 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1920 * config/cris/cris.h: Delete SDB reference in comment.
1921 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1922 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1923 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1924 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1926 (ix86_expand_prologue): Likewise.
1927 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1928 * config/ia64/ia64.h: Likewise.
1929 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1930 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1932 * config/mmix/mmix.h: Likewise.
1933 * config/nds32/nds32.c: Likewise.
1934 * config/stormy/storym16.h: Likewise.
1935 * config/visium/visium.h: Likewise.
1936 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1938 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1940 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1941 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1942 FRAME_POINTER_REGNUM point at high end of local var area.
1944 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1946 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1947 Move comment around. Do not reset best_edge for a copiable
1948 destination if the copy would cause a partition change.
1949 (better_edge_p): Remove redundant check.
1951 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1953 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1955 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1957 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1958 for math functions that have _Float<N> and _Float<N>X variants.
1959 (mathfn_built_in_2): Add support for math functions that have
1960 _Float<N> and _Float<N>X variants.
1961 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1962 (expand_builtin_mathfn_ternary): Add support for fma with
1963 _Float<N> and _Float<N>X variants.
1964 (expand_builtin): Likewise.
1965 (fold_builtin_3): Likewise.
1966 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1967 create math function _Float<N> and _Float<N>X variants as external
1969 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1970 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1971 the __builtin_ prefix and if not strict ansi, without the prefix.
1972 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1973 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1974 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1975 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1976 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1977 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1978 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1979 function signatures for fma _Float<N> and _Float<N>X variants.
1980 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1981 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1982 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1983 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1984 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1985 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1986 * gencfn-macros.c (print_case_cfn): Add support for math functions
1987 that have _Float<N> and _Float<N>X variants.
1988 (print_define_operator_list): Likewise.
1989 (fltfn_suffixes): Likewise.
1991 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1992 for math functions that have _Float<N> and _Float<N>X variants.
1993 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1994 and _Float<N>X variants.
1995 (COPYSIGN): Likewise.
1998 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1999 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2001 (integer_valued_read_call_p): Likewise.
2002 * fold-const-call.c (fold_const_call_ss): Likewise.
2003 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2004 _Float<N> and _Float<N>X variants.
2005 (fold_const_call_ssss): Add support for fma _Float<N> and
2006 _Float<N>X variants.
2007 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2008 support for copysign and fma _Float<N> and _Float<N>X variants.
2009 (backprop::process_builtin_call_use): Likewise.
2010 * tree-call-cdce.c (can_test_argument_range); Add support for
2011 sqrt _Float<N> and _Float<N>X variants.
2012 (edom_only_function): Likewise.
2013 (get_no_error_domain): Likewise.
2014 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2015 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2016 copysign _Float<N> and _Float<N>X variants.
2017 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2018 handled by machine independent code.
2019 (FMAF128): Likewise.
2020 * doc/cpp.texi (Common Predefined Macros): Document defining
2021 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2022 fma _Float<N> and _Float<N>X variants.
2024 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2027 * config/i386/i386-modes.def (CCFPU): Remove definition.
2028 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2029 (ix86_cc_modes_compatible): Ditto.
2030 (ix86_expand_carry_flag_compare): Ditto.
2031 (ix86_expand_int_movcc): Ditto.
2032 (ix86_expand_int_addcc): Ditto.
2033 (ix86_reverse_condition): Ditto.
2034 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2035 Return true/false for unordered/ordered fp comparisons.
2036 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2037 (ix86_prepare_fp_compare_args): Update for rename.
2038 (ix86_expand_fp_compare): Update for rename. Generate unordered
2039 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2040 (ix86_expand_sse_compare_and_jump): Ditto.
2041 * config/i386/predicates.md (fcmov_comparison_operator):
2042 Remove CCFPU mode handling.
2043 (ix86_comparison_operator): Ditto.
2044 (ix86_carry_flag_operator): Ditto.
2045 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2046 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2047 (*cmpu<mode>_cc_i387): Ditto.
2048 (FPCMP): Remove mode iterator.
2049 (unord): Remove mode attribute.
2050 (unord_subst): New define_subst transformation
2051 (unord): New define_subst attribute.
2053 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2054 (*cmpi<unord>xf_i387): Ditto.
2055 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2056 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2057 using unord_subst transformation.
2058 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2059 (round_saeonly): Also handle CCFP mode.
2060 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2061 Remove UNSPEC_SAHF unspec handling.
2063 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2065 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2067 2017-10-27 Jeff Law <law@redhat.com>
2069 * gimple-ssa-sprintf.c: Include domwalk.h.
2070 (class sprintf_dom_walker): New class, derived from dom_walker.
2071 (sprintf_dom_walker::before_dom_children): New function.
2072 (struct call_info): Moved into sprintf_dom_walker class
2073 (compute_formath_length, handle_gimple_call): Likewise.
2074 (sprintf_length::execute): Call the dominator walker rather
2075 than walking the statements.
2077 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2078 gimple statement locations.
2079 (check_array_bounds): Corresponding changes. Get the statement's
2080 location directly from wi->stmt.
2082 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2085 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2087 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2089 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2090 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2092 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2095 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2096 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2097 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2098 maybe_get_pool_constant.
2099 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2102 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2104 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2105 2.26 caveat. Update gas and gld versions.
2106 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2109 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2111 * cgraph.h (set_malloc_flag): Declare.
2112 * cgraph.c (set_malloc_flag_1): New function.
2113 (set_malloc_flag): Likewise.
2114 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2115 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2117 (read_ipa_call_summary): Add support for reading is_return_callee.
2118 (write_ipa_call_summary): Stream is_return_callee.
2119 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2120 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2121 ipa-prop.h, ipa-fnsummary.h.
2122 (pure_const_names): Change to static.
2123 (malloc_state_e): Define.
2124 (malloc_state_names): Define.
2125 (funct_state_d): Add field malloc_state.
2126 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2127 (check_retval_uses): New function.
2128 (malloc_candidate_p): Likewise.
2129 (analyze_function): Add support for malloc attribute.
2130 (pure_const_write_summary): Stream malloc_state.
2131 (pure_const_read_summary): Add support for reading malloc_state.
2132 (dump_malloc_lattice): New function.
2133 (propagate_malloc): New function.
2134 (warn_function_malloc): New function.
2135 (ipa_pure_const::execute): Call propagate_malloc and
2136 ipa_free_fn_summary.
2137 (pass_local_pure_const::execute): Add support for malloc attribute.
2138 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2139 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2141 2017-10-27 Martin Liska <mliska@suse.cz>
2143 PR gcov-profile/82457
2144 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2145 for fork-like functions to be properly instrumented.
2147 2017-10-27 Richard Biener <rguenther@suse.de>
2150 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2151 info when we redirected EH.
2153 2017-10-26 Michael Collison <michael.collison@arm.com>
2155 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2157 (<optab>_trunchf<GPI:mode>2: New pattern.
2158 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2159 * config/aarch64/iterators.md (wv): New mode attribute.
2160 (vf, VF): New mode attributes.
2161 (vgp, VGP): New mode attributes.
2162 (s): Update attribute with SImode and DImode prefixes.
2164 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2166 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2167 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2168 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2169 (nios2_option_overide): Initialize it. Don't allow R0-relative
2170 addressing with PIC.
2171 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2172 (nios2_symbolic_constant_p): Likewise.
2173 (nios2_legitimate_address_p): Likewise.
2174 (nios2_r0rel_section_name_p): New.
2175 (nios2_symbol_ref_in_r0rel_data_p): New.
2176 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2177 (r0rel_constant_p): New.
2178 (nios2_print_operand_address): Handle r0rel_constant_p.
2179 (nios2_cdx_narrow_form_p): Likewise.
2180 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2181 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2182 (Nios II Options): Document -mr0rel-sec.
2184 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2186 * config/nios2/nios2.c: Include xregex.h.
2187 (nios2_gprel_sec_regex): New.
2188 (nios2_option_overide): Initialize it. Don't allow GP-relative
2189 addressing with PIC.
2190 (nios2_small_section_name_p): Check for regex match.
2191 * config/nios2/nios2.opt (mgprel-sec=): New option.
2192 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2193 (Nios II Options): Document -mgprel-sec.
2195 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2197 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2199 2017-10-26 Tom de Vries <tom@codesourcery.com>
2201 PR tree-optimization/82707
2202 * gimple.c (gimple_copy): Fix unsharing of
2203 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2205 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2207 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2208 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2209 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2210 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2211 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2212 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2213 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2214 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2215 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2216 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2217 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2218 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2219 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2220 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2221 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2222 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2223 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2225 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2227 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2229 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2230 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2231 warning to rs6000.c. Remove the Undocumented flag, since it has
2233 (-mabi=ibmlongdouble): Likewise.
2234 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2235 already set, set the default format for long double.
2236 (rs6000_debug_reg_global): Print whether long double is IBM or
2238 (rs6000_option_override_internal): Rework setting long double
2239 format. Only warn if the user is changing the long double default
2240 and they did not use -Wno-psabi.
2241 * doc/invoke.texi (PowerPC options): Update the documentation for
2242 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2244 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2245 Alan Hayward <alan.hayward@arm.com>
2246 David Sherwood <david.sherwood@arm.com>
2248 * rtl.h (wider_subreg_mode): New function.
2249 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2250 rather than an unsigned int *.
2251 * ira-color.c (regno_max_ref_width): Replace with...
2252 (regno_max_ref_mode): ...this new variable.
2253 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2254 Use wider_subreg_mode.
2255 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2256 rather than an unsigned int *.
2257 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2258 (process_alt_operands): Likewise.
2259 (invariant_p): Likewise.
2260 * lra-spills.c (assign_mem_slot): Likewise.
2261 (add_pseudo_to_slot): Likewise.
2262 * lra.c (collect_non_operand_hard_regs): Likewise.
2263 (add_regs_to_insn_regno_info): Likewise.
2264 * reload1.c (regno_max_ref_width): Replace with...
2265 (regno_max_ref_mode): ...this new variable.
2266 (reload): Update accordingly. Update call to
2267 ira_sort_regnos_for_alter_reg.
2268 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2269 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2270 (scan_paradoxical_subregs): Likewise.
2272 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2274 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2275 (aarch64_frame): Add emit_frame_chain boolean.
2276 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2277 Move eh_return case to aarch64_layout_frame.
2278 (aarch64_layout_frame): Initialize emit_frame_chain.
2279 (aarch64_expand_prologue): Use emit_frame_chain.
2281 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2283 * config/aarch64/aarch64.c (aarch64_layout_frame):
2284 Ensure LR is always stored at the bottom of the callee-saves.
2285 Remove rarely used frame layout which saves callee-saves at top of
2286 frame, so the store of LR can be used as a valid probe in all cases.
2288 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2290 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2291 Improve unaligned TImode/TFmode base/offset split.
2293 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2294 Alan Hayward <alan.hayward@arm.com>
2295 David Sherwood <david.sherwood@arm.com>
2297 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2298 * combine.c (find_single_use_1): Likewise.
2299 (expand_field_assignment): Likewise.
2300 (move_deaths): Likewise.
2301 * lra-constraints.c (simplify_operand_subreg): Likewise.
2302 (curr_insn_transform): Likewise.
2303 * lra.c (collect_non_operand_hard_regs): Likewise.
2304 (add_regs_to_insn_regno_info): Likewise.
2305 * rtlanal.c (reg_referenced_p): Likewise.
2306 (covers_regno_no_parallel_p): Likewise.
2308 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2310 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2311 Don't print any bits outside the precision of the value.
2312 * wide-int.cc (test_printing): Add some new tests.
2314 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2316 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2317 supports -xbrace_comment option.
2318 * configure: Regenerate.
2319 * config.in: Regenerate.
2320 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2321 (ASM_CPU_SPEC): Use it.
2323 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2325 * target.def (static_rtx_alignment): New hook.
2326 * targhooks.h (default_static_rtx_alignment): Declare.
2327 * targhooks.c (default_static_rtx_alignment): New function.
2328 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2329 * doc/tm.texi: Regenerate.
2330 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2331 instead of targetm.constant_alignment. Remove call to
2333 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2334 (cris_preferred_mininum_alignment): New function, split out from...
2335 (cris_constant_alignment): ...here.
2336 (cris_static_rtx_alignment): New function.
2337 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2339 (ix86_constant_alignment): ...here.
2340 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2341 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2342 (mmix_static_rtx_alignment): New function.
2343 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2344 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2346 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2349 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2350 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2352 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2354 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2355 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2356 mark as a sign-extending load.
2357 (local_pic_load_u): Define.
2359 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2362 * fold-const.c (operand_equal_for_comparison_p): Also return true
2363 if ARG0 is a simple variant of ARG1 with narrower precision.
2364 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2366 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2368 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2370 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2371 scatter_static, scatter_per_elt.
2372 * x86-tune-costs.h: Add new cost entries.
2374 2017-10-25 Richard Biener <rguenther@suse.de>
2376 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2377 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2378 class pass_fre): Move to ...
2379 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2380 class pass_fre): ... here and adjust for statistics.
2382 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2385 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2386 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2387 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2388 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2391 2017-10-25 Richard Biener <rguenther@suse.de>
2393 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2394 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2395 eliminate_push_avail, eliminate_insert): Move inside...
2396 (class eliminate_dom_walker): ... this class in preparation
2398 (fini_eliminate): Remove by merging with ...
2399 (eliminate): ... this function. Adjust for class changes.
2400 (pass_pre::execute): Remove fini_eliminate call.
2401 (pass_fre::execute): Likewise.
2403 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2406 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2407 (VPERMI2, VPERMI2I): New mode iterators.
2408 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2409 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2411 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2412 mode iterator. Remove 3 old define_insn patterns.
2413 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2414 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2415 VPERMI2 mode iterator, remove the other two expanders.
2416 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2417 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2418 instructions, remove the other two patterns.
2419 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2420 mode iterator, remove the other two patterns.
2421 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2422 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2423 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2424 and adjust argument order accordingly.
2425 (ix86_expand_vec_perm): Adjust caller.
2426 (expand_vec_perm_1): Likewise.
2427 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2428 (expand_vec_perm_vpermt2_vpshub2): ... this.
2429 (ix86_expand_vec_perm_const_1): Adjust caller.
2430 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2433 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2434 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2435 (vec_shl_<mode>): Remove unused expander.
2436 (avx512bw_<shift_insn><mode>3): New define_insn.
2437 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2438 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2440 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2443 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2446 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2448 PR rtl-optimization/82396
2449 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2450 (autopref_multipass_init): Simplify initialization.
2451 (autopref_rank_data): Simplify sort order.
2452 * gcc/sched-int.h (autopref_multipass_data_): Remove
2453 multi_mem_insn_p, min_offset and max_offset.
2455 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2458 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2459 Check special value of flag_omit_frame_pointer.
2460 (aarch64_can_eliminate): Likewise.
2461 (aarch64_override_options_after_change_1): Simplify handling of
2462 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2464 2017-10-24 Richard Biener <rguenther@suse.de>
2466 PR tree-optimization/82697
2467 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2468 zero for conditional load and unconditional store.
2470 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2472 * doc/install.texi: Document bootstrap-cet.
2474 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2477 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2478 ENDBR instruction at function entrance if function is only
2481 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2484 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2485 patterns to better describe from which operation the CF is computed.
2486 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2487 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2488 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2489 is 0, use _0 suffixed expanders instead of emitting a comparison
2492 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2494 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2495 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2497 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2500 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2501 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2502 * loop-iv.c (iv_get_reaching_def): Likewise.
2503 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2504 variable is promoted and the partition contains undefined values.
2506 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2508 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2510 (nios2_address_cost): Define.
2511 (nios2_legitimize_address): Recognize (exp + constant) directly.
2512 (TARGET_ADDRESS_COST): Define.
2514 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2516 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2517 (nios2_symbolic_memory_operand_p): Declare.
2518 (nios2_split_large_constant): Declare.
2519 (nios2_split_symbolic_memory_operand): Declare.
2520 * config/nios2/nios2.c: Adjust includes.
2521 (nios2_symbolic_constant_allowed): New.
2522 (nios2_symbolic_constant_p): New.
2523 (nios2_plus_symbolic_constant_p): New.
2524 (nios2_valid_addr_expr_p): Recognize addresses involving
2526 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2527 (nios2_symbolic_memory_operand_p): New.
2528 (nios2_large_constant_p): New.
2529 (nios2_split_large_constant): New.
2530 (nios2_split_plus_large_constant): New.
2531 (nios2_split_symbolic_memory_operand): New.
2532 (nios2_legitimize_address): Code refactoring. Handle addresses
2533 involving symbolic constants.
2534 (nios2_emit_move_sequence): Likewise.
2535 (nios2_print_operand): Improve error output.
2536 (nios2_print_operand_address): Handle LO_SUM.
2537 (nios2_cdx_narrow_form_p): Likewise.
2538 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2539 operands involving symbolic constants.
2540 (movhi_internal, movsi_internal): Likewise.
2541 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2542 (extendhisi2, extendqi<mode>2): Likewise.
2544 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2546 * tree-pass.h (PROP_rtl_split_insns): Define.
2547 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2549 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2551 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2553 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2556 * target.def (const_not_ok_for_debug_p): Default to
2557 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2558 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2559 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2560 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2561 which targetm.const_not_ok_for_debug_p returned true.
2562 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2564 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2566 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2567 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2568 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2570 (ix86_const_not_ok_for_debug_p): New function.
2571 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2572 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2574 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2577 * system.h: Conditionally include "unique-ptr.h" if
2578 INCLUDE_UNIQUE_PTR is defined.
2579 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2580 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2582 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2584 * config/rl78/rl78.md: New define_expand "subdi3".
2586 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2589 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2590 DF_REF_INSN if DF_REF_INSN_INFO is false.
2592 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2594 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2595 xmm_move instead of sse_move.
2596 (sse_store_index): New function.
2597 (ix86_register_move_cost): Be more sensible about mismatch stall;
2598 model AVX moves correctly; make difference between sse->integer and
2600 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2601 moves; make difference between SSE and AVX.
2602 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2603 and zmm_move. Increase size of sse load and store tables;
2604 add unaligned load and store tables; add ssemmx_to_integer.
2605 * x86-tune-costs.h: Update all entries according to real
2606 move latencies from Agner Fog's manual and chip documentation.
2608 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2611 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2612 * config/i386/constraints.md (Wf): New constraint.
2613 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2614 (cmp<dwi>_doubleword): Removed.
2615 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2616 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2617 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2618 expand with cmp<dwi>_doubleword. For LTU and GEU use
2619 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2621 * common.opt (gcolumn-info): Enable by default.
2622 * doc/invoke.texi (gcolumn-info): Document new default.
2624 2017-10-23 Richard Biener <rguenther@suse.de>
2626 PR tree-optimization/82672
2627 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2628 Fold the stmt if we propagated into it.
2630 2017-10-23 Richard Biener <rguenther@suse.de>
2632 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2633 (bitmap_remove_expr_from_set): ... this. All callers call this
2634 for non-constant values.
2635 (bitmap_set_subtract): Rename to...
2636 (bitmap_set_subtract_expressions): ... this. Adjust and
2638 (bitmap_set_contains_value): Remove superfluous check.
2639 (bitmap_set_replace_value): Inline into single caller ...
2640 (bitmap_value_replace_in_set): ... here and simplify.
2641 (dependent_clean): Merge into ...
2642 (clean): ... this using an overload. Adjust.
2643 (prune_clobbered_mems): Adjust.
2644 (compute_antic_aux): Likewise.
2645 (compute_partial_antic_aux): Likewise.
2647 2017-10-23 Richard Biener <rguenther@suse.de>
2649 PR tree-optimization/82129
2651 2017-08-01 Richard Biener <rguenther@suse.de>
2653 PR tree-optimization/81181
2654 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2655 (compute_antic): ... end of iteration here.
2657 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2659 * target.def (starting_frame_offset): New hook.
2660 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2661 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2662 * doc/tm.texi.in: Regenerate.
2663 * hooks.h (hook_hwi_void_0): Declare.
2664 * hooks.c (hook_hwi_void_0): New function.
2665 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2666 STARTING_FRAME_OFFSET.
2667 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2668 * reload1.c (reload): Likewise.
2669 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2670 instead of STARTING_FRAME_OFFSET.
2671 * function.c (try_fit_stack_local): Likewise.
2672 (assign_stack_local_1): Likewise
2673 (instantiate_virtual_regs): Likewise.
2674 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2675 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2676 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2677 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2678 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2679 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2680 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2681 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2682 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2683 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2684 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2685 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2686 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2687 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2688 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2689 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2690 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2691 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2692 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2693 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2694 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2695 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2696 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2697 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2698 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2699 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2700 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2701 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2702 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2703 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2704 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2705 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2706 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2707 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2708 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2709 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2710 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2711 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2712 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2713 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2714 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2715 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2716 return a HOST_WIDE_INT.
2717 (avr_builtin_setjmp_frame_value): Use it instead of
2718 STARTING_FRAME_OFFSET.
2719 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2720 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2721 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2723 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2724 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2725 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2726 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2727 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2728 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2729 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2730 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2731 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2732 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2733 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2734 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2736 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2737 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2738 * config/mips/mips.c (mips_compute_frame_info): Refer to
2739 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2740 (mips_starting_frame_offset): New function.
2741 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2742 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2743 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2744 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2745 and return a HOST_WIDE_INT.
2746 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2747 (mmix_initial_elimination_offset): Refer to
2748 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2749 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2750 * config/pa/pa.c (pa_starting_frame_offset): New function.
2751 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2752 (pa_expand_prologue): Likewise.
2753 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2754 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2755 !FRAME_GROWS_DOWNWARD handling to...
2756 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2757 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2758 !FRAME_GROWS_DOWNWARD handling to...
2759 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2760 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2761 !FRAME_GROWS_DOWNWARD handling to...
2762 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2763 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2765 (rs6000_starting_frame_offset): New function.
2766 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2767 !FRAME_GROWS_DOWNWARD handling to...
2768 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2769 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2770 !FRAME_GROWS_DOWNWARD handling to...
2771 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2772 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2773 !FRAME_GROWS_DOWNWARD handling to...
2774 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2775 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2776 (rs6000_starting_frame_offset): New function.
2777 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2778 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2779 * config/vax/vax.c (vax_starting_frame_offset): New function.
2780 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2781 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2782 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2783 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2784 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2785 * system.h (STARTING_FRAME_OFFSET): Poison.
2787 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2789 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2790 SCALAR_TYPE_MODE instead of TYPE_MODE.
2792 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2793 Alan Hayward <alan.hayward@arm.com>
2794 David Sherwood <david.sherwood@arm.com>
2796 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2798 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2799 Alan Hayward <alan.hayward@arm.com>
2800 David Sherwood <david.sherwood@arm.com>
2802 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2804 2017-10-23 Richard Biener <rguenther@suse.de>
2806 PR tree-optimization/82129
2807 * tree-ssa-pre.c (bitmap_set_and): Remove.
2808 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2809 canonicalizing expressions in the set to those with lowest
2810 ID rather than taking that from the first edge.
2812 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2814 * combine.c (rtx_equal_for_field_assignment_p): Use
2815 byte_lowpart_offset.
2817 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2818 Alan Hayward <alan.hayward@arm.com>
2819 David Sherwood <david.sherwood@arm.com>
2821 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2822 to a SUBREG_PROMOTED_VAR.
2824 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2825 Alan Hayward <alan.hayward@arm.com>
2826 David Sherwood <david.sherwood@arm.com>
2828 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2829 (expand_debug_source_expr): Likewise.
2830 * combine.c (combine_simplify_rtx): Likewise.
2831 * cse.c (fold_rtx): Likewise.
2832 * optabs.c (expand_float): Likewise.
2833 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2834 (simplify_binary_operation_1): Likewise.
2836 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2837 Alan Hayward <alan.hayward@arm.com>
2838 David Sherwood <david.sherwood@arm.com>
2840 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2841 (record_promoted_value): Likewise.
2842 * expr.c (expand_expr_real_2): Likewise.
2843 * ree.c (update_reg_equal_equiv_notes): Likewise.
2844 (combine_set_extension): Likewise.
2845 * rtlanal.c (low_bitmask_len): Likewise.
2846 * simplify-rtx.c (neg_const_int): Likewise.
2847 (simplify_binary_operation_1): Likewise.
2849 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2850 Alan Hayward <alan.hayward@arm.com>
2851 David Sherwood <david.sherwood@arm.com>
2853 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2854 * regcprop.c (maybe_mode_change): Likewise.
2855 * reload1.c (alter_reg): Likewise.
2857 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2859 * inchash.h (inchash::hash::add_wide_int): New function.
2860 * lto-streamer-out.c (hash_tree): Use it.
2862 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2864 * inchash.h (inchash::hash::add_wide_int): Rename to...
2865 (inchash::hash::add_hwi): ...this.
2866 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2867 (polymorphic_call_target_hasher::hash): Likewise.
2868 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2869 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2870 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2871 * lto-streamer-out.c (hash_tree): Likewise.
2872 * optc-save-gen.awk: Likewise.
2873 * tree.c (add_expr): Likewise.
2875 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2878 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2879 for ordered inequality comparisons even with TARGET_IEEE_FP.
2881 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2884 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2885 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2886 Expand with cmp<dwi>_doubleword.
2888 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2890 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2891 List CET intrinsics.
2892 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2893 specific to -fcf-protection option.
2895 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2897 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2898 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2899 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2900 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2901 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2902 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2903 (extra_objs): Add cet.o for Linux/x86 targets.
2904 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2905 * config/i386/cet.c: New file.
2906 * config/i386/cetintrin.h: Likewise.
2907 * config/i386/t-cet: Likewise.
2908 * config/i386/cpuid.h (bit_SHSTK): New.
2909 (bit_IBT): Likewise.
2910 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2911 pass IBT and SHSTK bits.
2912 * config/i386/i386-builtin-types.def
2913 (VOID_FTYPE_UNSIGNED_PVOID): New.
2914 (VOID_FTYPE_UINT64_PVOID): Likewise.
2915 * config/i386/i386-builtin.def: Add CET intrinsics.
2916 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2917 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2918 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2919 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2921 * config/i386/i386.c (rest_of_insert_endbranch): New.
2922 (pass_data_insert_endbranch): Likewise.
2923 (pass_insert_endbranch): Likewise.
2924 (make_pass_insert_endbranch): Likewise.
2925 (ix86_notrack_prefixed_insn_p): Likewise.
2926 (ix86_target_string): Add -mibt, -mshstk flags.
2927 (ix86_option_override_internal): Add flag_cf_protection
2929 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2930 (ix86_print_operand): Add 'notrack' prefix output.
2931 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2932 (ix86_expand_builtin): Expand CET intrinsics.
2933 (x86_output_mi_thunk): Add 'endbranch' instruction.
2934 * config/i386/i386.h (TARGET_IBT): New.
2935 (TARGET_IBT_P): Likewise.
2936 (TARGET_SHSTK): Likewise.
2937 (TARGET_SHSTK_P): Likewise.
2938 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2939 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2940 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2941 (builtin_setjmp_setup): New pattern.
2942 (builtin_longjmp): Likewise.
2943 (rdssp<mode>): Likewise.
2944 (incssp<mode>): Likewise.
2945 (saveprevssp): Likewise.
2946 (rstorssp): Likewise.
2947 (wrss<mode>): Likewise.
2948 (wruss<mode>): Likewise.
2949 (setssbsy): Likewise.
2950 (clrssbsy): Likewise.
2951 (nop_endbr): Likewise.
2952 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2954 * config/i386/immintrin.h: Include <cetintrin.h>.
2955 * config/i386/linux-common.h
2956 (file_end_indicate_exec_stack_and_cet): New prototype.
2957 (TARGET_ASM_FILE_END): New.
2959 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2961 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2962 latencies instead of having separate table; make difference between
2963 integer and float costs.
2964 * i386.h (processor_costs): Remove scalar_stmt_cost,
2965 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2966 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2968 * x86-tune-costs.h: Remove entries which has been removed in
2969 procesor_costs from all tables; make cond_taken_branch_cost
2970 and cond_not_taken_branch_cost COST_N_INSNS based.
2972 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2974 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2976 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2978 * config/i386/i386.md (isa): Remove fma_avx512f.
2979 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2980 <avx512>_fmadd_<mode>_mask3<round_name>,
2981 <avx512>_fmsub_<mode>_mask<round_name>,
2982 <avx512>_fmsub_<mode>_mask3<round_name>,
2983 <avx512>_fnmadd_<mode>_mask<round_name>,
2984 <avx512>_fnmadd_<mode>_mask3<round_name>,
2985 <avx512>_fnmsub_<mode>_mask<round_name>,
2986 <avx512>_fnmsub_<mode>_mask3<round_name>,
2987 <avx512>_fmaddsub_<mode>_mask<round_name>,
2988 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2989 <avx512>_fmsubadd_<mode>_mask<round_name>,
2990 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2991 (*vec_widen_umult_even_v16si<mask_name>,
2992 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2993 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2995 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2997 * extend.texi: Add 'nocf_check' documentation.
2998 * gimple.texi: Add second parameter to
2999 gimple_build_call_from_tree.
3000 * invoke.texi: Add -fcf-protection documentation.
3001 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3003 2017-10-20 Richard Biener <rguenther@suse.de>
3005 PR tree-optimization/82473
3006 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3007 the largest input type.
3009 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3011 * c-attribs.c (handle_nocf_check_attribute): New function.
3012 (c_common_attribute_table): Add 'nocf_check' handling.
3013 * gimple-parser.c: Add second argument NULL to
3014 gimple_build_call_from_tree.
3015 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3016 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3018 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3019 * common.opt: Add fcf-protection flag.
3020 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3021 * flag-types.h: Add enum cf_protection_level.
3022 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3023 Add 'nocf_check' attribute propagation to gimple call.
3024 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3025 (gimple_build_call_from_tree): Update prototype.
3026 (gimple_call_nocf_check_p): New function.
3027 (gimple_call_set_nocf_check): Likewise.
3028 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3029 * ipa-icf.c: Add nocf_check attribute in statement hash.
3030 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3031 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3032 * toplev.c (process_options): Add flag_cf_protection handling.
3034 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3036 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3038 2017-10-20 Richard Biener <rguenther@suse.de>
3040 PR tree-optimization/82603
3041 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3042 remove false predicated stores.
3044 2017-10-20 Richard Biener <rguenther@suse.de>
3046 * graphite-isl-ast-to-gimple.c
3047 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3048 Remove return value and simplify, dump copied stmt after lhs
3050 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3051 Reduce dump verbosity.
3052 (gsi_insert_earliest): Likewise.
3053 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3054 * graphite.c (print_global_statistics): Adjust dumping.
3055 (print_graphite_scop_statistics): Likewise.
3056 (print_graphite_statistics): Do not dump loops here.
3057 (graphite_transform_loops): But here.
3059 2017-10-20 Nicolas Roche <roche@adacore.com>
3061 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3062 * configure: Regenerate.
3064 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3067 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3068 functions when optimizing replace GIMPLE_RETURN stmts with
3069 calls to __builtin_unreachable ().
3072 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3073 for -fsanitize=thread link of executables.
3074 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3075 link of executables.
3078 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3080 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3081 define_insns for logical vector shifts to use VI248_AVX512BW
3082 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3083 condition, useless isa and prefix attributes. Change the first
3084 2 of these define_insns to ...
3085 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3086 define_insn for avx512vl.
3087 (<shift_insn><mode>3): ... and this, new define_insn without
3088 masking for non-avx512vl.
3091 * config/i386/sse.md (*andnot<mode>3,
3092 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3093 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3094 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3095 not applied use empty suffix even for TARGET_AVX512VL.
3096 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3097 is applied, supply evex,evex or evex,evex,evex instead of just
3100 2017-10-20 Julia Koval <julia.koval@intel.com>
3102 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3103 (OPTION_MASK_ISA_GFNI_UNSET): New.
3104 (ix86_handle_option): Handle OPT_mgfni.
3105 * config/i386/cpuid.h (bit_GFNI): New.
3106 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3107 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3108 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3109 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3110 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3111 * config/i386/i386.opt: Add mgfni.
3113 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3115 * config/msp430/msp430.c (msp430_option_override): Disable
3116 -fdelete-null-pointer-checks.
3117 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3119 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3121 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3122 of x87 and SSE instructions.
3124 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3126 * asan.c (create_cond_insert_point): Do not update edge count.
3127 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3128 (afdo_propagate_circuit): Likewise.
3129 (afdo_calculate_branch_prob): Likewise.
3130 (afdo_annotate_cfg): Likewise.
3131 * basic-block.h (struct edge_def): Remove count.
3132 (edge_def::count): New accessor.
3133 * bb-reorder.c (rotate_loop): Update.
3134 (find_traces_1_round): Update.
3135 (connect_traces): Update.
3136 (sanitize_hot_paths): Update.
3137 * cfg.c (unchecked_make_edge): Update.
3138 (make_single_succ_edge): Update.
3139 (check_bb_profile): Update.
3140 (dump_edge_info): Update.
3141 (update_bb_profile_for_threading): Update.
3142 (scale_bbs_frequencies_int): Update.
3143 (scale_bbs_frequencies_gcov_type): Update.
3144 (scale_bbs_frequencies_profile_count): Update.
3145 (scale_bbs_frequencies): Update.
3146 * cfganal.c (connect_infinite_loops_to_exit): Update.
3147 * cfgbuild.c (compute_outgoing_frequencies): Update.
3148 (find_many_sub_basic_blocks): Update.
3149 * cfgcleanup.c (try_forward_edges): Update.
3150 (try_crossjump_to_edge): Update
3151 * cfgexpand.c (expand_gimple_cond): Update
3152 (expand_gimple_tailcall): Update
3153 (construct_exit_block): Update
3154 * cfghooks.c (verify_flow_info): Update
3155 (redirect_edge_succ_nodup): Update
3156 (split_edge): Update
3157 (make_forwarder_block): Update
3158 (duplicate_block): Update
3159 (account_profile_record): Update
3160 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3161 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3162 * cfgloopmanip.c (scale_loop_profile): Update.
3164 (lv_adjust_loop_entry_edge): Update.
3165 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3166 (force_nonfallthru_and_redirect): Update.
3167 (purge_dead_edges): Update.
3168 (rtl_flow_call_edges_add): Update.
3169 * cgraphunit.c (init_lowered_empty_function): Update.
3170 (cgraph_node::expand_thunk): Update.
3171 * gimple-pretty-print.c (dump_probability): Update.
3172 (dump_edge_probability): Update.
3173 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3174 * haifa-sched.c (sched_create_recovery_edges): Update.
3175 * hsa-gen.c (convert_switch_statements): Update.
3176 * ifcvt.c (dead_or_predicable): Update.
3177 * ipa-inline-transform.c (inline_transform): Update.
3178 * ipa-split.c (split_function): Update.
3179 * ipa-utils.c (ipa_merge_profiles): Update.
3180 * loop-doloop.c (add_test): Update.
3181 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3182 * lto-streamer-in.c (input_cfg): Update.
3183 (input_function): Update.
3184 * lto-streamer-out.c (output_cfg): Update.
3185 * modulo-sched.c (sms_schedule): Update.
3186 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3187 * predict.c (maybe_hot_edge_p): Update.
3188 (unlikely_executed_edge_p): Update.
3189 (probably_never_executed_edge_p): Update.
3190 (dump_prediction): Update.
3191 (drop_profile): Update.
3192 (propagate_unlikely_bbs_forward): Update.
3193 (determine_unlikely_bbs): Update.
3194 (force_edge_cold): Update.
3195 * profile.c (compute_branch_probabilities): Update.
3196 * reg-stack.c (better_edge): Update.
3197 * shrink-wrap.c (handle_simple_exit): Update.
3198 * tracer.c (better_p): Update.
3199 * trans-mem.c (expand_transaction): Update.
3200 (split_bb_make_tm_edge): Update.
3201 * tree-call-cdce.c: Update.
3202 * tree-cfg.c (gimple_find_sub_bbs): Update.
3203 (gimple_split_edge): Update.
3204 (gimple_duplicate_sese_region): Update.
3205 (gimple_duplicate_sese_tail): Update.
3206 (gimple_flow_call_edges_add): Update.
3207 (insert_cond_bb): Update.
3208 (execute_fixup_cfg): Update.
3209 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3210 * tree-complex.c (expand_complex_div_wide): Update.
3211 * tree-eh.c (lower_resx): Update.
3212 (unsplit_eh): Update.
3213 (cleanup_empty_eh_move_lp): Update.
3214 * tree-inline.c (copy_edges_for_bb): Update.
3215 (freqs_to_counts): Update.
3216 (copy_cfg_body): Update.
3217 * tree-ssa-dce.c (remove_dead_stmt): Update.
3218 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3219 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3220 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3221 (unloop_loops): Update.
3222 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3223 * tree-ssa-loop-split.c (connect_loops): Update.
3224 (split_loop): Update.
3225 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3226 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3227 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3228 * tree-ssa-reassoc.c (branch_fixup): Update.
3229 * tree-ssa-tail-merge.c (replace_block_by): Update.
3230 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3231 (compute_path_counts): Update.
3232 (update_profile): Update.
3233 (recompute_probabilities): Update.
3234 (update_joiner_offpath_counts): Update.
3235 (estimated_freqs_path): Update.
3236 (freqs_to_counts_path): Update.
3237 (clear_counts_path): Update.
3238 (ssa_fix_duplicate_block_edges): Update.
3239 (duplicate_thread_path): Update.
3240 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3241 (case_bit_test_cmp): Update.
3242 (collect_switch_conv_info): Update.
3243 (gen_inbound_check): Update.
3244 (do_jump_if_equal): Update.
3245 (emit_cmp_and_jump_insns): Update.
3246 * tree-tailcall.c (decrease_profile): Update.
3247 (eliminate_tail_call): Update.
3248 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3249 (vect_do_peeling): Update.
3250 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3251 * ubsan.c (ubsan_expand_null_ifn): Update.
3252 (ubsan_expand_ptr_ifn): Update.
3253 * value-prof.c (gimple_divmod_fixed_value): Update.
3254 (gimple_mod_pow2): Update.
3255 (gimple_mod_subtract): Update.
3256 (gimple_ic): Update.
3257 (gimple_stringop_fixed_value): Update.
3259 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3262 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3264 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3266 PR rtl-optimization/82395
3267 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3268 based on non_spilled_static_chain_regno_p.
3270 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3272 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3273 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3274 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3276 2017-10-19 Martin Sebor <msebor@redhat.com>
3278 PR tree-optimization/82596
3279 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3281 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3283 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3284 (get_mem_refs_of_builtin_call): Likewise.
3285 * builtins.c (expand_builtin_apply): Adjust call to
3286 allocate_dynamic_stack_space.
3287 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3288 the third argument to allocate_dynamic_stack_space, otherwise -1.
3289 (expand_builtin): Deal with all alloca variants.
3290 (is_inexpensive_builtin): Likewise.
3291 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3292 * calls.c (special_function_p): Deal with all alloca variants.
3293 (initialize_argument_information): Adjust call to
3294 allocate_dynamic_stack_space.
3295 (expand_call): Likewise.
3296 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3297 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3298 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3299 use it for the stack usage computation.
3300 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3301 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3302 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3303 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3304 (in_loop_p): Remove first argument and useless check.
3305 (pass_walloca::execute): Remove useless test and adjust call to above.
3306 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3307 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3308 (gimplify_call_expr): Deal with all alloca variants.
3309 * hsa-gen.c (gen_hsa_alloca): Likewise.
3310 (gen_hsa_insns_for_call): Likewise.
3311 * ipa-pure-const.c (special_builtin_state): Likewise.
3312 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3313 * tree-object-size.c (alloc_object_size): Likewise.
3314 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3315 (call_may_clobber_ref_p_1): Likewise.
3316 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3317 (ccp_fold_stmt): Likewise.
3318 (optimize_stack_restore): Likewise.
3319 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3320 (mark_all_reaching_defs_necessary_1): Likewise.
3321 (propagate_necessity): Likewise.
3322 (eliminate_unnecessary_stmts): Likewise.
3323 * tree.c (build_common_builtin_nodes): Build
3324 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3325 (build_alloca_call_expr): New function.
3326 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3327 (CASE_BUILT_IN_ALLOCA): Likewise.
3328 (build_alloca_call_expr): Declare.
3329 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3331 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3334 * dwarf2out.c (new_die_raw): New static inline function.
3335 (new_die): Use it to create the DIE.
3336 (add_AT_external_die_ref): Likewise.
3337 (clone_die): Likewise.
3338 (clone_as_declaration): Likewise.
3339 (dwarf2out_vms_debug_main_pointer): Likewise.
3340 (base_type_die): Likewise. Remove early return for corner cases.
3341 Do not call add_pubtype on the DIE here.
3342 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3343 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3344 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3345 native order exists for base types, attach the DIE manually and call
3346 add_pubtype on it. Do not equate a reverse order DIE to the type.
3348 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3350 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3351 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3353 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3355 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3357 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3358 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3359 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3360 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3361 * opts.c (sanitizer_opts): Add builtin.
3362 * ubsan.c (instrument_builtin): New function.
3363 (pass_ubsan::execute): Call it.
3364 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3365 * doc/invoke.texi: Document -fsanitize=builtin.
3367 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3368 builtins, store max (log2 (align), 0) into uchar field instead of
3369 align into uptr field.
3370 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3371 store uchar 0 field instead of uptr 0 field.
3372 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3373 instead of passing one address of struct with 2 locations pass
3374 two addresses of structs with 1 location each.
3375 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3376 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3377 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3378 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3379 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3380 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3381 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3382 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3384 2017-10-19 Martin Liska <mliska@suse.cz>
3387 * file-find.c (remove_prefix): Remove.
3388 * file-find.h (remove_prefix): Likewise.
3389 * gcc-ar.c: Remove smartness of lookup.
3391 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3393 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3394 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3395 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3397 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3400 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3401 (setcc + and to xor + setcc): New peephole2.
3403 2017-10-19 Tom de Vries <tom@codesourcery.com>
3405 * doc/sourcebuild.texi (Test Directives, Variants of
3406 dg-require-support): Add dg-require-stack-size.
3408 2017-10-19 Martin Liska <mliska@suse.cz>
3411 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3412 that have a large alignment.
3413 (gimplify_target_expr): Likewise.
3415 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3417 PR rtl-optimization/82602
3418 * ira.c (rtx_moveable_p): Return false for volatile asm.
3420 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3423 * config/i386/i386-modes.def (CCGZ): New CC mode.
3424 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3425 * config/i386/predicates.md (ix86_comparison_operator):
3427 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3428 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3429 with double-word subtraction.
3430 (put_condition_code): Handle CCGZmode.
3432 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3434 * wide-int.cc (debug (const wide_int &)): New.
3435 (debug (const wide_int *)): New.
3436 (debug (const widest_int &)): New.
3437 (debug (const widest_int *)): New.
3439 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3442 * lra-constraints.c (curr_insn_transform): Use non-input operand
3443 instead of output one for matched reload.
3445 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3447 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3448 (tree-ssa-loop-ivopts.h): New header file.
3449 (struct builtin_info): New fields.
3450 (classify_builtin_1): Compute and record base and offset parts for
3451 memset builtin partition by calling strip_offset.
3452 (offset_cmp, fuse_memset_builtins): New functions.
3453 (finalize_partitions): Fuse adjacent memset partitions by calling
3455 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3456 Expose the interface.
3457 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3459 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3461 PR tree-optimization/82574
3462 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3463 that data reference must be executed exactly once per iteration
3464 against the outermost loop in nest.
3465 (classify_partition): Update call to above function.
3467 2017-10-18 Richard Biener <rguenther@suse.de>
3469 PR tree-optimization/82591
3470 * graphite.c (graphite_transform_loops): Move code gen message
3472 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3473 Here. Handle scop_to_isl_ast failing.
3474 (scop_to_isl_ast): Limit the number of ISL operations.
3476 2017-10-18 Richard Biener <rguenther@suse.de>
3478 * graphite-isl-ast-to-gimple.c
3479 (translate_isl_ast_to_gimple::set_rename): Simplify.
3480 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3481 (graphite_copy_stmts_from_block): ... here.
3482 (copy_bb_and_scalar_dependences): Simplify.
3483 (add_parameters_to_ivs_params): Canonicalize.
3484 (generate_entry_out_of_ssa_copies): Simplify.
3485 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3486 by passing in ISL dimension.
3487 (parameter_index_in_region_1): Rename to ...
3488 (parameter_index_in_region): ... this.
3489 (extract_affine): Adjust assert, pass down parameter index.
3490 (add_param_constraints): Use range-info when available.
3491 (build_scop_context): Adjust.
3492 * sese.c (new_sese_info): Adjust.
3493 (free_sese_info): Likewise.
3494 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3495 Remove unused typedefs.
3496 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3498 2017-10-18 Martin Liska <mliska@suse.cz>
3500 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3502 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3504 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3505 (s390_sched_init): Do not reset s390_sched_state if we entered the
3506 current basic block via a fallthru edge and all others are unlikely.
3508 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3510 * config/s390/s390.c (NUM_SIDES): New variable.
3511 (LONGRUNNING_THRESHOLD): New variable.
3512 (LATENCY_FACTOR): New variable.
3513 (s390_sched_score): Decrease score for long-running instructions on
3515 (s390_sched_variable_issue): Perform bookkeeping for long-running
3518 2017-10-18 Richard Biener <rguenther@suse.de>
3520 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3521 Simplify with removal of the parameter rename map.
3522 (set_rename): Likewise.
3523 (should_copy_to_new_region): Likewise.
3524 (graphite_copy_stmts_from_block): Likewise.
3525 (copy_bb_and_scalar_dependences): Remove initialization of
3526 unused copied_bb_map.
3528 (copy_internal_parameters): Likewise.
3529 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3530 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3531 Use INTEGRAL_TYPE_P.
3532 (parameter_index_in_region_1): Rename to ...
3533 (assign_parameter_index_in_region): ... this. Assert we have
3534 a parameter we handle.
3535 (scan_tree_for_params): Adjust.
3536 * sese.h (parameter_rename_map_t): Remove.
3537 (struct sese_info_t): Remove unused parameter_rename_map and
3538 copied_bb_map members.
3539 * sese.c (new_sese_info): Adjust.
3540 (free_sese_info): Likewise.
3542 2017-10-18 Martin Liska <mliska@suse.cz>
3545 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3546 on an abnormal edge.
3548 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3550 * doc/invoke.texi (ffunction-sections and fdata-sections):
3553 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3555 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3556 the use statement can throw internally.
3558 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3560 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3561 any RTX present on the RHS of a SET.
3562 * compare-elim.c (try_eliminate_compare): Restore comment.
3564 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3566 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3567 may be also called on tcc_type nodes.
3568 * langhooks.c (lhd_tree_size): Likewise.
3570 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3572 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3573 format_warning_at_substring.
3574 (maybe_warn): Convert source_range * param to a location_t. Pass
3575 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3576 (format_directive): Remove code to extract source_ranges and
3577 source_range * in favor of just a location_t.
3578 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3580 * substring-locations.c (format_warning_va): Convert
3581 source_range * param to a location_t.
3582 (format_warning_at_substring): Likewise.
3583 * substring-locations.h (format_warning_va): Likewise.
3584 (format_warning_at_substring): Likewise.
3586 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3588 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3590 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3591 and scatter/gather ops.
3593 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3594 vec_gather_load and vec_scatter_store.
3595 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3596 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3597 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3598 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3599 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3600 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3602 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3604 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3605 Detect FTST instruction and handle its register pops. Only pop
3606 second operand if can_pop_second_op is true.
3607 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3608 set can_pop_second_op to false in the compare_for_stack_reg call.
3610 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3611 output_fp_compare for stack register operands.
3612 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3613 instructions here. Do not emit stack register pops here. Assert
3614 that FCOMPP pops next to top stack register. Rewrite function.
3616 2017-10-17 Nathan Sidwell <nathan@acm.org>
3619 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3620 use DECL_ASSEMBLER_NAME_RAW.
3623 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3626 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3627 Wilco Dijkstra <wilco.dijkstra@arm.com>
3629 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3630 converstion to Pmode from the buf_addr.
3632 2017-10-17 Richard Biener <rguenther@suse.de>
3634 * graphite-dependences.c (scop_get_reads_and_writes): Change
3635 output parameters to references.
3637 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3639 PR 71026/tree-optimization
3640 * fold-const.c (distribute_real_division): Removed.
3641 (fold_binary_loc): Remove calls to distribute_real_divison.
3643 2017-10-17 Richard Biener <rguenther@suse.de>
3645 * graphite-scop-detection.c
3646 (scop_detection::stmt_has_simple_data_refs_p): Always use
3647 the full nest as region.
3648 (try_generate_gimple_bb): Likewise.
3649 * sese.c (scalar_evolution_in_region): Simplify now that
3650 SCEV can handle instantiation in regions.
3651 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3652 in the non-loop part of a function if requested.
3654 2017-10-17 Richard Biener <rguenther@suse.de>
3656 PR tree-optimization/82563
3657 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3659 (graphite_regenerate_ast_isl): Call it.
3660 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3662 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3664 PR tree-optimization/82549
3665 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3666 Formatting fixes. Instead of calling make_bit_field_ref with negative
3669 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3671 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3672 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3673 _mm_maskz_reduce_ss): New.
3674 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3675 __builtin_ia32_reducess_mask): Ditto..
3676 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3677 * config/i386/sse.md (reduces<mode>): Renamed to ...
3678 (reduces<mode><mask_scalar_name>): ... this.
3679 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3681 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3682 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3684 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3686 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3687 * selftest-run-tests.c (selftest::run_tests): Call
3688 selftest::unique_ptr_tests_cc_tests.
3689 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3690 * unique-ptr-tests.cc: New file.
3692 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3695 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3697 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3698 (make_hard_regno_born, make_hard_regno_dead): Update
3699 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3701 2017-10-16 Jeff Law <law@redhat.com>
3703 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3705 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3707 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3709 2017-10-16 Olivier Hainque <hainque@adacore.com>
3711 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3712 with_cpu if we were configured for an e500v2 target cpu name.
3714 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3716 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3717 * doc/invoke.texi: Document +nodsp as a valid extension for
3720 2017-10-16 Martin Liska <mliska@suse.cz>
3722 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3723 (test_set_range): Likewise.
3724 (test_range_functions): Rename to ...
3725 (test_bit_in_range): ... this.
3726 (sbitmap_c_tests): Add new test.
3728 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3730 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3732 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3733 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3735 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3737 * config/aarch64/aarch64-builtins.c
3738 (aarch64_types_quadopu_lane_qualifiers): New.
3739 (TYPES_QUADOPU_LANE): New.
3740 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3741 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3742 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3743 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3744 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3745 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3746 (Vdottype, DOTPROD): New.
3747 (sur): Add SDOT and UDOT.
3749 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3751 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3752 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3753 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3755 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3756 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3757 Enable TARGET_DOTPROD.
3758 (cortex-a75.cortex-a55): Likewise.
3759 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3761 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3763 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3764 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3766 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3768 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3769 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3770 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3771 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3772 * config/arm/types.md (neon_dot, neon_dot_q): New.
3773 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3775 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3777 * config/arm/arm.h (TARGET_DOTPROD): New.
3778 * config/arm/arm.c (arm_arch_dotprod): New.
3779 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3780 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3781 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3782 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3783 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3784 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3785 * doc/invoke.texi (armv8.2-a): Document dotprod
3787 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3789 * i386.c (ix86_vec_cost): New function.
3790 (ix86_rtx_costs): Handle vector operations better.
3791 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3792 * x86-tune-costs.h: Add new costs to all tables.
3794 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3796 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3798 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3799 divsd, sqrtss and sqrtsd
3800 * x86-tune-costs.h: Add new entries to all costs.
3801 (znver1_cost): Fix to match real instruction latencies.
3803 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3804 Michael Collison <michael.collison@arm.com>
3806 * compare-elim.c: Include emit-rtl.h.
3807 (can_merge_compare_into_arith): New function.
3808 (try_validate_parallel): Likewise.
3809 (try_merge_compare): Likewise.
3810 (try_eliminate_compare): Call the above when no previous clobber
3812 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3815 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3819 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3820 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3824 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3825 to be any operand_equal_p operands. For & (B - 1) require
3826 B to be power of 2. Recognize
3827 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3829 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3832 * optabs.c (expand_memory_blockage): Fix call of
3833 targetm.have_memory_blockage.
3835 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3838 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3839 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3840 objects to extra_objs instead of overwriting it.
3842 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3844 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3845 Use any_fp_register_operand as operand[3] predicate. Simplify
3846 equality test for operands[2] and operands[4] memory location.
3847 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3848 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3849 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3850 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3851 any_fp_register_operand as operand[1] predicate. Simplify
3852 equality test for operands[0] and operands[3] memory location.
3853 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3854 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3855 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3857 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3859 * target-insns.def: Add memory_blockage.
3860 * optabs.c (expand_memory_blockage): New function.
3861 (expand_asm_memory_barrier): Rename ...
3862 (expand_asm_memory_blockage): ... to this.
3863 (expand_mem_thread_fence): Call expand_memory_blockage
3864 instead of expand_asm_memory_barrier.
3865 (expand_mem_singnal_fence): Ditto.
3866 (expand_atomic_load): Ditto.
3867 (expand_atomic_store): Ditto.
3868 * doc/md.texi (Standard Pattern Names For Generation):
3869 Document memory_blockage instruction pattern.
3871 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3873 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3874 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3875 * config/rl78/rl78.md: New define_expand "adddi3".
3877 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3879 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3882 2017-10-13 Jeff Law <law@redhat.com>
3884 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3886 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3889 * internal-fn.c (expand_mul_overflow): If both operands have
3890 the same highpart of -1 or 0 and the topmost bit of lowpart
3891 is different, overflow is if res <= 0 rather than res < 0.
3893 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3895 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3896 TARGET_P9_VECTOR code for unaligned_load case.
3898 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3900 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3902 2017-10-13 Nathan Sidwell <nathan@acm.org>
3904 * tree-core.h (tree_contains_struct): Make bool.
3905 * tree.c (tree_contains_struct): Likewise.
3906 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3907 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3908 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3909 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3910 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3912 2017-10-13 Richard Biener <rguenther@suse.de>
3914 * graphite-isl-ast-to-gimple.c
3915 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3916 parameters and dominance check.
3917 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3918 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3919 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3920 Do not update SSA form here or do intermediate IL verification.
3921 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3922 (graphite_initialize): Remove check on the number of loops in
3923 the function and inline into graphite_transform_loops.
3924 (graphite_finalize): Inline into graphite_transform_loops.
3925 (graphite_transform_loops): Perform SSA update and IL verification
3927 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3929 2017-10-13 Richard Biener <rguenther@suse.de>
3931 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3932 graphite_expression_type_precision): Avoid global constructor
3934 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3935 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3936 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3937 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3938 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3939 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3941 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3944 * config/i386/i386.h (ix86_red_zone_size): New.
3945 * config/i386/i386.md (push peephole2s): Replace
3946 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3948 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3949 Alan Hayward <alan.hayward@arm.com>
3950 David Sherwood <david.sherwood@arm.com>
3952 * combine.c (can_change_dest_mode): Reject changes in
3953 REGMODE_NATURAL_SIZE.
3955 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3956 Alan Hayward <alan.hayward@arm.com>
3957 David Sherwood <david.sherwood@arm.com>
3959 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3960 (expand_debug_source_expr): Likewise.
3961 * combine.c (combine_simplify_rtx): Likewise.
3962 * cse.c (fold_rtx): Likewise.
3963 * fwprop.c (canonicalize_address): Likewise.
3964 * targhooks.c (default_shift_truncation_mask): Likewise.
3966 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3967 Alan Hayward <alan.hayward@arm.com>
3968 David Sherwood <david.sherwood@arm.com>
3970 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3971 (widened_mode): Likewise.
3972 (expand_unop): Likewise.
3973 * ree.c (transform_ifelse): Likewise.
3974 (merge_def_and_ext): Likewise.
3975 (combine_reaching_defs): Likewise.
3976 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3978 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3979 Alan Hayward <alan.hayward@arm.com>
3980 David Sherwood <david.sherwood@arm.com>
3982 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3983 * combine.c (gen_lowpart_for_combine): Likewise.
3984 * dwarf2out.c (rtl_for_decl_location): Likewise.
3985 * final.c (alter_subreg): Likewise.
3986 * rtlhooks.c (gen_lowpart_general): Likewise.
3987 (gen_lowpart_if_possible): Likewise.
3989 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3990 Alan Hayward <alan.hayward@arm.com>
3991 David Sherwood <david.sherwood@arm.com>
3993 * calls.c (expand_call): Use subreg_lowpart_offset.
3994 * cse.c (cse_insn): Likewise.
3995 * regcprop.c (copy_value): Likewise.
3996 (copyprop_hardreg_forward_1): Likewise.
3998 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4001 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4002 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4003 =Q constraints to +Q and into insn condition add check
4004 that operands[0] and operands[1] are equal.
4005 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4006 =Q constraints to +Q and into insn condition add check
4007 that operands[0] is equal to either operands[1] or operands[2].
4010 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4011 instead of handling MINUS_EXPR twice (once for each argument),
4012 canonicalize operand order and handle just once, use rtype where
4013 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4016 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4017 any values of __C while still being pattern recognizable as a simple
4020 2017-10-13 Richard Biener <rguenther@suse.de>
4022 PR tree-optimization/82451
4024 2017-10-02 Richard Biener <rguenther@suse.de>
4026 PR tree-optimization/82355
4027 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4028 a mapping for the enclosing loop but avoid generating one for
4030 (copy_bb_and_scalar_dependences): Remove premature codegen
4031 error on PHIs in blocks duplicated into multiple places.
4032 * graphite-scop-detection.c
4033 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4034 in the region use it as loop and nest to analyze the DR in.
4035 (try_generate_gimple_bb): Likewise.
4036 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4037 (add_loop_constraints): For blocks in a loop not in the region
4038 create a dimension with a single iteration.
4039 * sese.h (gbb_loop_at_index): Remove assert.
4041 * cfgloop.c (loop_preheader_edge): For the loop tree root
4042 return the single successor of the entry block.
4043 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4044 Reset the SCEV hashtable and niters.
4045 * graphite-scop-detection.c
4046 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4047 assert that we only have POLYNOMIAL_CHREC that vary in loops
4048 contained in the region.
4049 (scop_detection::graphite_can_represent_expr): Adjust.
4050 (scop_detection::stmt_has_simple_data_refs_p): For loops
4051 not in the region set loop to NULL. The nest is now the
4052 entry edge to the region.
4053 (try_generate_gimple_bb): Likewise.
4054 * sese.c (scalar_evolution_in_region): Adjust for
4055 instantiate_scev change.
4056 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4057 Make nest parameter the edge into the region.
4058 (create_data_ref): Likewise.
4059 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4060 entry edge into a region and adjust instantiate_scev calls.
4061 (create_data_ref): Likewise.
4062 (graphite_find_data_references_in_stmt): Likewise.
4063 (find_data_references_in_stmt): Pass the loop preheader edge
4064 from the nest argument.
4065 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4066 parameter the edge into the region.
4067 (instantiate_parameters): Use the loop preheader edge as entry.
4068 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4070 (get_instantiated_value_entry): Make instantiate_below parameter
4071 the edge into the region.
4072 (instantiate_scev_name): Likewise. Adjust dominance checks,
4073 when we cannot use loop-based instantiation instantiate by
4074 walking use-def chains.
4075 (instantiate_scev_poly): Adjust.
4076 (instantiate_scev_binary): Likewise.
4077 (instantiate_scev_convert): Likewise.
4078 (instantiate_scev_not): Likewise.
4079 (instantiate_array_ref): Remove.
4080 (instantiate_scev_3): Likewise.
4081 (instantiate_scev_2): Likewise.
4082 (instantiate_scev_1): Likewise.
4083 (instantiate_scev_r): Do not blindly handle N-operand trees.
4084 Do not instantiate array-refs. Handle all constants and invariants.
4085 (instantiate_scev): Make instantiate_below parameter
4086 the edge into the region.
4087 (resolve_mixers): Use the loop preheader edge for the region
4088 parameter to instantiate_scev_r.
4089 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4091 2017-10-13 Richard Biener <rguenther@suse.de>
4093 PR tree-optimization/82525
4094 * graphite-isl-ast-to-gimple.c
4095 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4097 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4098 Fail code generation when we cannot represent the isl integer.
4099 (binary_op_to_tree): Elide modulo operations that are no-ops
4100 in the type we code generate. Remove now superfluous code
4103 2017-10-13 Richard Biener <rguenther@suse.de>
4105 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4106 (scop_detection::harmful_loop_in_region): Remove premature
4107 IV type restriction.
4108 (scop_detection::graphite_can_represent_scev): We can handle
4109 pointer IVs just fine.
4111 2017-10-13 Alan Modra <amodra@gmail.com>
4113 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4114 "Clobbers and Scratch Registers". Add paragraph on
4115 alternative to clobbers for scratch registers and OpenBLAS
4118 2017-10-13 Alan Modra <amodra@gmail.com>
4120 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4121 example of a memory input for a string of known length. Move
4122 commentary out of table. Add a number of new examples
4123 covering array memory inputs.
4125 2017-10-12 Martin Liska <mliska@suse.cz>
4127 PR tree-optimization/82493
4128 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4129 (test_range_functions): New function.
4130 (sbitmap_c_tests): Likewise.
4131 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4132 * selftest.h (sbitmap_c_tests): New function.
4134 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4136 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4138 * config/rs6000/amo.h: Fix spacing issue.
4140 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4143 * config/i386/i386.md (*ashl<mode>3_mask_1,
4144 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4145 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4148 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4150 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4151 (profile_probability): Set max_probability
4152 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4154 * profile-count.c (profile_probability::differs_from_p): Do not
4155 rely on max_probaiblity == 10000
4157 2017-10-12 Jeff Law <law@redhat.com>
4159 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4162 2017-10-12 Martin Sebor <msebor@redhat.com>
4166 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4167 (handle_alias_pairs): Call it.
4168 * common.opt (-Wattribute-alias): New option.
4169 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4170 * doc/invoke.texi (-Wattribute-alias): Document.
4172 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4175 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4177 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4179 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4180 (make_hard_regno_born, make_hard_regno_dead): Update
4181 bb_killed_pseudos and bb_gen_pseudos.
4183 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4185 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4187 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4189 * config/alpha/alpha.c (alpha_split_conditional_move):
4190 Use std::swap instead of manually swapping.
4191 (alpha_stdarg_optimize_hook): Ditto.
4192 (alpha_canonicalize_comparison): Ditto.
4194 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4196 * tree-loop-distribution.c (struct builtin_info): New struct.
4197 (struct partition): Refactor fields into struct builtin_info.
4198 (partition_free): Free struct builtin_info.
4199 (build_size_arg_loc, build_addr_arg_loc): Delete.
4200 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4201 information from struct builtin_info.
4202 (find_single_drs): New function refactored from classify_partition.
4203 Also moved builtin validity checks to this function.
4204 (compute_access_range, alloc_builtin): New functions.
4205 (classify_builtin_st, classify_builtin_ldst): New functions.
4206 (classify_partition): Refactor code into functions find_single_drs,
4207 classify_builtin_st and classify_builtin_ldst.
4208 (distribute_loop): Don't do runtime alias check when distributing
4210 (find_seed_stmts_for_distribution): New function.
4211 (pass_loop_distribution::execute): Refactor code finding seed
4212 stmts into above function. Support distribution for the innermost
4213 two-level loop nest. Adjust dump information.
4215 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4217 * tree-loop-distribution.c: Adjust the general comment.
4218 (NUM_PARTITION_THRESHOLD): New macro.
4219 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4220 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4221 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4222 in call to build_partition_graph.
4223 (finalize_partitions): New parameter. Make loop distribution more
4224 conservative by fusing more partitions.
4225 (distribute_loop): Don't do runtime alias check in case of loop nest
4227 (find_seed_stmts_for_distribution): New function.
4228 (prepare_perfect_loop_nest): New function.
4229 (pass_loop_distribution::execute): Refactor code finding seed stmts
4230 and loop nest into above functions. Support loop nest distribution.
4231 Adjust dump information accordingly.
4233 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4235 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4236 and set PTYPE_SEQUENTIAL for merged partition.
4238 2017-10-12 Richard Biener <rguenther@suse.de>
4240 PR tree-optimization/69728
4242 2017-09-19 Richard Biener <rguenther@suse.de>
4244 PR tree-optimization/69728
4245 * graphite-sese-to-poly.c (schedule_error): New global.
4246 (add_loop_schedule): Handle empty domain by failing the
4248 (build_original_schedule): Handle schedule_error.
4250 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4251 domain by returning an unchanged schedule.
4253 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4255 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4256 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4258 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4260 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4263 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4266 * expr.c (store_field): Don't optimize away bitsize == 0 store
4267 from CALL_EXPR with addressable return type.
4269 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4271 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4272 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4273 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4274 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4275 TARGET_ISEL instead of TARGET_ISEL<sel>.
4277 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4279 * config/rs6000/rs6000.c
4280 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4282 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4284 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4286 (reg_or_cint_operand): Fix comment.
4287 (reg_or_zero_operand): New predicate.
4288 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4289 * config/rs6000/rs6000.c (output_isel): Delete.
4290 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4291 instead of reg_or_cint_operand. Output instruction directly (not via
4293 (isel_unsigned_<mode>): Ditto.
4294 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4295 gpc_reg_operand. Add an instruction alternative for this. Output
4296 instruction directly.
4297 (*isel_reversed_unsigned_<mode>): Ditto.
4299 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4301 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4302 (TARGET_CANONICALIZE_COMPARISON): Define.
4304 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4307 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4308 Check whether the dest is REG before adding REG_EQUIV note.
4310 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4313 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4315 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4316 (make_hard_regno_born, make_hard_regno_dead): Update
4317 bb_killed_pseudos and bb_gen_pseudos.
4319 2017-10-11 Nathan Sidwell <nathan@acm.org>
4321 * incpath.h (enum incpath_kind): Name enum, prefix values.
4322 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4323 * incpath.c (heads, tails): Use INC_MAX.
4324 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4325 (merge_include_chains, split_quote_chain,
4326 register_include_chains): Update incpath_kind names.
4327 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4328 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4330 (add_framework_path, darwin_register_objc_includes): Likewise.
4331 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4333 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4335 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4336 Do not use float_operator operator predicate.
4337 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4338 * config/i386/predicates.md (float_operator): Remove predicate.
4340 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4342 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4343 (*jccxf_i387): Ditto.
4344 (*jcc<mode>_i387): Ditto.
4345 (*jccu<mode>_i387): Ditto.
4346 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4347 (*jcc_*_i387 splitters): Remove.
4348 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4349 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4350 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4353 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4355 * profile-count.h (slow_safe_scale_64bit): New function.
4356 (safe_scale_64bit): New inline.
4357 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4358 * profile-count.c: Include wide-int.h
4359 (slow_safe_scale_64bit): New.
4361 2017-10-11 Nathan Sidwell <nathan@acm.org>
4363 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4364 HAS_DECL_ASSEMBLER_NAME_P.
4365 * gimple-expr.c (gimple_decl_printable_name: Check
4366 HAS_DECL_ASSEMBLER_NAME_P too.
4367 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4368 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4369 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4370 * passes.c (rest_of_decl_compilation): Check
4371 HAS_DECL_ASSEMBLER_NAME_P too.
4372 * recog.c (verify_changes): Likewise.
4373 * tree-pretty-print.c (dump_decl_name): Likewise.
4374 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4376 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4377 (SET_DECL_ASSEMBLER_NAME): Use it.
4378 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4379 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4380 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4382 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4384 * config.gcc (i386, x86_64): Add extra objects.
4385 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4386 (ix86_min_insn_size): Declare.
4387 (ix86_issue_rate): Declare.
4388 (ix86_adjust_cost): Declare.
4389 (ia32_multipass_dfa_lookahead): Declare.
4390 (ix86_macro_fusion_p): Declare.
4391 (ix86_macro_fusion_pair_p): Declare.
4392 (ix86_bd_has_dispatch): Declare.
4393 (ix86_bd_do_dispatch): Declare.
4394 (ix86_core2i7_init_hooks): Declare.
4395 (ix86_atom_sched_reorder): Declare.
4396 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4397 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4398 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4399 (rip_relative_addr_p): Rename to ...
4400 (ix86_rip_relative_addr_p): ... this one; export.
4401 (memory_address_length): Update.
4402 (ix86_issue_rate): Move to x86-tune-sched.c.
4403 (ix86_flags_dependent): Move to x86-tune-sched.c.
4404 (ix86_agi_dependent): Move to x86-tune-sched.c.
4405 (exact_dependency_1): Move to x86-tune-sched.c.
4406 (exact_store_load_dependency): Move to x86-tune-sched.c.
4407 (ix86_adjust_cost): Move to x86-tune-sched.c.
4408 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4409 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4410 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4411 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4412 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4413 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4414 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4415 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4416 (min_insn_size): Rename to ...
4417 (ix86_min_insn_size): ... this one; export.
4418 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4419 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4420 (core2i7_first_cycle_multipass_backtrack): Move to
4421 x86-tune-sched-core.c.
4422 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4423 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4424 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4425 (ix86_avoid_jump_mispredicts): Update.
4426 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4427 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4428 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4429 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4430 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4431 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4432 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4433 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4434 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4435 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4436 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4437 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4438 (BIG): Move to ix86-tune-sched-bd.c.
4439 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4440 (enum insn_path): Move to ix86-tune-sched-bd.c.
4441 (get_mem_group): Move to ix86-tune-sched-bd.c.
4442 (is_cmp): Move to ix86-tune-sched-bd.c.
4443 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4444 (is_branch): Move to ix86-tune-sched-bd.c.
4445 (is_prefetch): Move to ix86-tune-sched-bd.c.
4446 (init_window): Move to ix86-tune-sched-bd.c.
4447 (allocate_window): Move to ix86-tune-sched-bd.c.
4448 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4449 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4450 (process_end_window): Move to ix86-tune-sched-bd.c.
4451 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4452 (find_constant): Move to ix86-tune-sched-bd.c.
4453 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4454 (has_immediate): Move to ix86-tune-sched-bd.c.
4455 (get_insn_path): Move to ix86-tune-sched-bd.c.
4456 (get_insn_group): Move to ix86-tune-sched-bd.c.
4457 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4458 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4459 (add_insn_window): Move to ix86-tune-sched-bd.c.
4460 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4461 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4462 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4463 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4464 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4465 (do_dispatch): Move to ix86-tune-sched-bd.c.
4466 (has_dispatch): Move to ix86-tune-sched-bd.c.
4467 * config/i386/t-i386: Add new object files.
4468 * config/i386/x86-tune-costs.h: New file.
4469 * config/i386/x86-tune-sched-atom.c: New file.
4470 * config/i386/x86-tune-sched-bd.c: New file.
4471 * config/i386/x86-tune-sched-core.c: New file.
4472 * config/i386/x86-tune-sched.c: New file.
4474 2017-10-11 Liu Hao <lh_mouse@126.com>
4476 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4477 the generic version below instead.
4478 (should_colorize): Recognize Windows consoles as terminals
4480 * pretty-print.c [__MINGW32__] (write_all): New function.
4481 [__MINGW32__] (find_esc_head): Likewise.
4482 [__MINGW32__] (find_esc_terminator): Likewise.
4483 [__MINGW32__] (eat_esc_sequence): Likewise.
4484 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4486 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4489 2017-10-11 Richard Biener <rguenther@suse.de>
4491 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4492 Properly call analyze_scalar_evolution with the loop of the stmt.
4494 2017-10-11 Richard Biener <rguenther@suse.de>
4496 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4497 * tree-core.h (tree_base): Add chrec_var union member.
4498 * tree.h (CHREC_VAR): Remove.
4499 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4500 * tree-chrec.h (build_polynomial_chrec): Adjust.
4501 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4502 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4504 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4506 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4507 * match.pd: ... here.
4508 ((T) X == (T) Y): Relax condition.
4510 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4512 PR tree-optimization/82472
4513 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4515 (break_alias_scc_partitions): Update postorder number.
4517 2017-10-11 Martin Liska <mliska@suse.cz>
4520 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4522 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4524 2017-10-11 Martin Liska <mliska@suse.cz>
4526 * print-rtl.c (print_insn): Move declaration of idbuf
4527 to same scope as name.
4529 2017-10-11 Martin Liska <mliska@suse.cz>
4534 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4537 2017-10-11 Martin Liska <mliska@suse.cz>
4540 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4543 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4545 * config.gcc (default_gnu_indirect_function): Default to yes for
4546 arm*-*-linux* with glibc.
4548 2017-10-11 Richard Biener <rguenther@suse.de>
4550 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4551 default-defs and types we do not want to analyze.
4552 (interpret_loop_phi): Replace unreachable code with an assert.
4553 (compute_scalar_evolution_in_loop): Remove and inline ...
4554 (analyze_scalar_evolution_1): ... here, replacing condition with
4555 what makes the intent clearer. Remove handling of cases
4556 get_scalar_evolution now handles.
4558 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4560 PR rtl-optimization/81434
4561 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4562 comment for main loop. In sched_group_found if, also add checks for
4563 pass and min_cost_group.
4565 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4567 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4568 (rs6000_insn_cost): New function.
4569 * config/rs6000/rs6000.md (cost): New attribute.
4571 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4572 H.J. Lu <hongjiu.lu@intel.com>
4576 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4577 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4578 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4579 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4580 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4581 OPTION_MASK_ISA_MMX and some other bit set require both
4582 mmx and the ISAs without the mmx bit.
4583 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4584 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4585 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4586 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4587 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4588 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4589 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4590 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4591 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4592 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4593 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4594 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4595 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4596 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4597 Add OPTION_MASK_ISA_MMX.
4599 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4601 * config.gcc (armv7*-*-freebsd*): New target.
4602 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4604 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4606 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4607 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4610 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4612 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4613 a template parameter.
4614 (WIDE_INT_REF_FOR): Update accordingly.
4615 * tree.h (wi::int_traits <const_tree>): Delete.
4616 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4617 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4618 (wi::tree_to_wide_ref): New typedef.
4619 (wi::to_wide): New function.
4620 * calls.c (get_size_range): Use wi::to_wide when operating on
4622 * cgraph.c (cgraph_node::create_thunk): Likewise.
4623 * config/i386/i386.c (ix86_data_alignment): Likewise.
4624 (ix86_local_alignment): Likewise.
4625 * dbxout.c (stabstr_O): Likewise.
4626 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4627 * expr.c (const_vector_from_tree): Likewise.
4628 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4629 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4630 (fold_negate_expr_1, int_const_binop_1, const_binop)
4631 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4632 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4633 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4634 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4635 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4636 (fold_not_const, round_up_loc): Likewise.
4637 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4638 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4639 (alloca_call_type): Likewise.
4640 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4641 * godump.c (go_output_typedef): Likewise.
4642 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4643 * internal-fn.c (get_min_precision): Likewise.
4644 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4645 * ipa-polymorphic-call.c
4646 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4647 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4648 (ipa_modify_call_arguments): Likewise.
4649 * match.pd: Likewise.
4650 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4651 * print-tree.c (print_node_brief, print_node): Likewise.
4652 * stmt.c (expand_case): Likewise.
4653 * stor-layout.c (layout_type): Likewise.
4654 * tree-affine.c (tree_to_aff_combination): Likewise.
4655 * tree-cfg.c (group_case_labels_stmt): Likewise.
4656 * tree-data-ref.c (dr_analyze_indices): Likewise.
4657 (prune_runtime_alias_test_list): Likewise.
4658 * tree-dump.c (dequeue_and_dump): Likewise.
4659 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4660 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4661 * tree-pretty-print.c (dump_generic_node): Likewise.
4662 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4663 (simple_iv_with_niters): Likewise.
4664 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4665 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4666 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4667 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4668 (refine_value_range_using_guard, number_of_iterations_ne_max)
4669 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4670 (get_cst_init_from_scev, record_nonwrapping_iv)
4671 (scev_var_range_cant_overflow): Likewise.
4672 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4673 * tree-ssa-pre.c (compute_avail): Likewise.
4674 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4675 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4676 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4677 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4678 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4679 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4680 (dump_case_nodes, try_switch_expansion): Likewise.
4681 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4682 (vect_do_peeling): Likewise.
4683 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4684 * tree-vect-stmts.c (vectorizable_load): Likewise.
4685 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4686 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4687 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4688 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4689 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4690 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4691 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4692 (evrp_dom_walker::before_dom_children): Likewise.
4693 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4694 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4695 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4696 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4697 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4698 (get_range_pos_neg): Likewise.
4699 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4700 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4701 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4702 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4703 * config/avr/avr.c (avr_fold_builtin): Likewise.
4704 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4705 * config/msp430/msp430.c (msp430_attr): Likewise.
4706 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4707 * config/powerpcspe/powerpcspe-c.c
4708 (altivec_resolve_overloaded_builtin): Likewise.
4709 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4710 (rs6000_expand_ternop_builtin): Likewise.
4711 * config/rs6000/rs6000-c.c
4712 (altivec_resolve_overloaded_builtin): Likewise.
4713 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4714 (rs6000_expand_ternop_builtin): Likewise.
4715 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4717 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4719 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4720 when copying loop nest with only one inner loop.
4722 2017-10-10 Richard Biener <rguenther@suse.de>
4724 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4725 blocks if SCEV is active.
4726 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4728 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4729 (scev_initialize): Assert we are not yet initialized.
4731 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4733 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4734 inner loop's exit stmt by making it always exit the loop, otherwise
4735 we would generate an infinite empty loop.
4737 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4739 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4740 renaming variables in new preheader if it's deleted.
4742 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4744 * tree-loop-distribution.c (struct partition): Remove unused field
4745 loops of the structure.
4746 (partition_alloc, partition_free): Ditto.
4747 (build_rdg_partition_for_vertex): Ditto.
4749 2017-10-09 Jeff Law <law@redhat.com>
4751 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4752 return type to match prototype and documentation.
4754 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4756 * config/rs6000/rs6000.c (processor_costs): Move to ...
4757 * config/rs6000/rs6000.h: ... here.
4758 (rs6000_cost): Declare.
4760 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4762 * except.c (setjmp_fn): New global variable.
4763 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4764 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4765 if DONT_USE_BUILTIN_SETJMP is defined.
4767 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4769 * target.def (insn_cost): New hook.
4770 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4771 * doc/tm.texi: Regenerate.
4772 * rtlanal.c (insn_cost): Use the new hook.
4774 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4776 * combine.c (combine_validate_cost): Compute the new insn_cost,
4777 not just pattern_cost.
4778 (try_combine): Adjust comment.
4780 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4782 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4784 * combine.c (uid_insn_cost): Adjust comment.
4785 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4787 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4788 * dse.c (find_shift_sequence): Ditto.
4789 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4790 (bb_valid_for_noce_process_p): Use pattern_cost.
4791 * rtl.h (insn_rtx_cost): Delete.
4792 (pattern_cost): New prototype.
4793 (insn_cost): New prototype.
4794 * rtlanal.c (insn_rtx_cost): Rename to...
4795 (pattern_cost): ... this.
4798 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4800 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4801 (*jcc<mode>_0_r_i387): Ditto.
4802 (*jccxf_r_i387): Ditto.
4803 (*jcc<mode>_r_i387): Ditto.
4804 (*jccu<mode>_r_i387): Ditto.
4805 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4806 (*jcc): Rename from *jcc_1.
4808 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4810 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4811 deferred rescans after the lvx/stvx recombination pre-pass.
4813 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4815 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4816 memory operation instruction support.
4817 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4818 (rs6000-ibm-aix[789]*): Likewise.
4819 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4820 Document new functions.
4822 2017-10-09 Richard Biener <rguenther@suse.de>
4824 PR tree-optimization/82397
4825 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4826 equality only for semantically equal trees.
4828 2017-10-09 Richard Biener <rguenther@suse.de>
4830 PR tree-optimization/82449
4831 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4832 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4833 allow constant addresses.
4834 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4837 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4839 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4842 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4845 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4848 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4851 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4853 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4856 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4857 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4859 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4861 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4862 (WI_BINARY_PREDICATE_RESULT): Likewise.
4863 (wi::binary_traits::operator_result): New type.
4864 (wi::binary_traits::predicate_result): Likewise.
4865 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4866 (generic_wide_int::operator==, generic_wide_int::operator!=)
4867 (generic_wide_int::operator&, generic_wide_int::and_not)
4868 (generic_wide_int::operator|, generic_wide_int::or_not)
4869 (generic_wide_int::operator^, generic_wide_int::operator+
4870 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4872 (operator~, unary operator-, operator==, operator!=, operator&)
4873 (operator|, operator^, operator+, binary operator-, operator*): New
4875 * expr.c (get_inner_reference): Use wi::bit_and_not.
4876 * fold-const.c (fold_binary_loc): Likewise.
4877 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4878 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4879 (bit_value_binop): Likewise.
4880 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4881 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4882 (extract_range_from_binary_expr_1): Likewise.
4883 (masked_increment): Likewise.
4884 (simplify_bit_ops_using_ranges): Likewise.
4886 2017-10-09 Martin Jambor <mjambor@suse.cz>
4889 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4890 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4891 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4893 (hsa_fixup_mov_insn_type): New function.
4894 (hsa_op_with_type::get_in_type): Use it.
4895 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4896 immediates in an assert.
4897 (hsa_op_with_type::extend_int_to_32bit): New method.
4898 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4899 types. Convert to dest type if necessary.
4900 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4901 (reg_for_gimple_ssa): Pass false as min32int to
4902 hsa_type_for_scalar_tree_type.
4903 (gen_hsa_addr): Fixup type when creating addresable temporary.
4904 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4905 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4906 necessary. Call hsa_fixup_mov_insn_type.
4907 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4908 extend operands and convert to dest type if necessary.
4909 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4910 to dest type if necessary.
4911 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4912 if conversion nt necessary and size matches.
4913 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4914 to dest type if necessary.
4915 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4916 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4918 (gen_hsa_clrsb): Likewise.
4919 (gen_hsa_ffs): Likewise.
4920 (gen_hsa_divmod): Extend operands and convert to dest type if
4922 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4924 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4926 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4927 Remove empty default arguments. Use a brace block as output
4929 (conditional return): Ditto.
4931 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4932 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4934 (group_ending_nop): Ditto.
4935 (doloop_end): Ditto.
4936 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4937 (splitters for those): Ditto.
4939 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4941 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4942 a conditional jump (and the compare for it) so that pc_rtx is the
4944 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4945 for the deleted and renamed ctr<mode>_internal[234] patterns.
4946 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4947 Delete second conditional return pattern.
4948 (ctr<mode>_internal2): Delete this second bdnz pattern.
4949 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4950 (ctr<mode>_internal4): Delete this second bdz pattern.
4952 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4954 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4955 (always_initialized_rtx_for_ssa_name_p): New predicate.
4956 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4957 (finish_out_of_ssa): Free new field of SA.
4958 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4959 * tree-ssa-coalesce.c: Include tree-ssa.h.
4960 (get_parm_default_def_partitions): Remove extern keyword.
4961 (get_undefined_value_partitions): New function.
4962 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4963 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4965 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4967 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4969 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4971 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4973 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4974 for targets that preffer 128bit.
4976 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4978 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4980 2017-10-08 Olivier Hainque <hainque@adacore.com>
4982 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4983 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4985 (thumb_set_return_address): Likewise.
4987 2017-10-08 Olivier Hainque <hainque@adacore.com>
4989 * common/config/arm/arm-common.c (arm_except_unwind_info):
4990 Handle DWARF2_UNWIND_INFO.
4992 2017-10-07 Michael Collison <michael.collison@arm.com>
4994 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4997 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4999 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5000 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5001 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5002 defined, force the creation of a new block for a dispatch label.
5004 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5006 * invoke.texi (Wsuggest-attribute=cold): Document.
5007 * common.opt (Wsuggest-attribute=cold): New
5008 * ipa-pure-const.c (warn_function_cold): New function.
5009 * predict.c (compute_function_frequency): Use it.
5010 * predict.h (warn_function_cold): Declare.
5012 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5014 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5017 2017-10-06 Martin Liska <mliska@suse.cz>
5019 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5020 keyword for member functions.
5021 (struct sanopt_tree_couple): New struct.
5022 (struct sanopt_tree_couple_hash): New function.
5023 (struct sanopt_ctx): Add new hash_map.
5024 (has_dominating_ubsan_ptr_check): New function.
5025 (record_ubsan_ptr_check_stmt): Likewise.
5026 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5027 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5028 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5030 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5033 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5034 aarch64_simd_valid_immediate on CONST_VECTORs.
5035 (aarch64_reg_or_bic_imm): Likewise.
5037 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5039 PR rtl-optimization/82396
5040 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5042 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5044 * graphite-dependences.c (scop_get_reads): Move code to...
5045 (scop_get_must_writes): Move code to...
5046 (scop_get_may_writes): Move code to...
5047 (scop_get_reads_and_writes): ... here.
5048 (scop_get_dependences): Call scop_get_reads_and_writes.
5050 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5052 PR tree-optimization/82434
5053 * fold-const.h (can_native_encode_type_p,
5054 can_native_encode_string_p): Remove.
5055 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5056 don't encode anything, just return what would be otherwise returned.
5057 (native_encode_fixed, native_encode_complex, native_encode_vector):
5059 (native_encode_string): Likewise. Inline by hand
5060 can_native_encode_string_p.
5061 (can_native_encode_type_p): Remove.
5062 (can_native_encode_string_p): Remove.
5063 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5064 STRING_CSTs using can_native_encode_string_p, test all
5065 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5066 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5067 argument from native_encode_expr.
5068 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5069 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5072 2017-10-06 Richard Biener <rguenther@suse.de>
5074 PR tree-optimization/82397
5075 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5076 operand_equal_p but rely on data_ref_compare_tree for detecting
5078 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5079 to match up with dr_group_sort_cmp.
5081 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5084 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5086 * config/s390/s390-builtin-types.def: Regenerate.
5088 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5091 * config/s390/s390-builtin-types.def: Regenerate.
5092 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5093 Change flag from B_VXE to B_VX.
5094 (s390_vec_min_dbl): Remove B_VXE flag.
5096 2017-10-06 Richard Biener <rguenther@suse.de>
5098 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5099 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5100 translate_isl_ast_to_gimple::is_valid_rename,
5101 translate_isl_ast_to_gimple::get_rename,
5102 translate_isl_ast_to_gimple::get_def_bb_for_const,
5103 translate_isl_ast_to_gimple::get_new_name,
5104 translate_isl_ast_to_gimple::collect_all_ssa_names,
5105 translate_isl_ast_to_gimple::copy_loop_phi_args,
5106 translate_isl_ast_to_gimple::collect_all_ssa_names,
5107 translate_isl_ast_to_gimple::copy_loop_phi_args,
5108 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5109 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5110 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5111 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5112 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5113 translate_isl_ast_to_gimple::copy_cond_phi_args,
5114 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5115 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5116 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5117 translate_isl_ast_to_gimple::rename_uses,
5118 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5119 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5120 (set_rename_for_each_def): Likewise.
5121 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5122 here. Handle rewriting SCEV analyzable uses here.
5123 (copy_bb_and_scalar_dependences): Generate code for PHI
5125 (graphite_regenerate_ast_isl): Adjust.
5126 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5127 (add_write, add_read): New functions.
5128 (build_cross_bb_scalars_def): Use it and simplify.
5129 (build_cross_bb_scalars_use): Likewise.
5130 (graphite_find_cross_bb_scalar_vars): Inline into...
5131 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5132 simulating out-of-SSA. Compute liveout and add dependencies.
5133 (build_scops): Force an empty entry block.
5134 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5136 (sese_build_liveouts): Declare.
5137 (sese_trivially_empty_bb_p): Likewise.
5138 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5139 compute liveout and debug_liveout.
5140 (sese_bad_liveouts_use): Remove.
5141 (sese_reset_debug_liveouts_bb): Likewise.
5142 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5143 (sese_build_liveouts): Build liveout and debug_liveout and store
5145 (new_sese_info): Adjust.
5146 (free_sese_info): Likewise.
5147 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5148 do not build liveout here.
5149 (move_sese_in_condition): Adjust region entry.
5150 (scev_analyzable_p): Match up with chrec_apply requirements.
5151 (sese_trivially_empty_bb_p): New.
5152 * tree-into-ssa.c (get_reaching_def): Properly support generating
5153 default-defs for incremental rewrite of anonymous names.
5155 2017-10-06 Richard Biener <rguenther@suse.de>
5157 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5158 precision do not perform modulo reduction.
5160 2017-10-06 Richard Biener <rguenther@suse.de>
5162 PR tree-optimization/82436
5163 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5164 conservatively choose the vectorization factor when checking
5165 whether we can perform the required load permutation.
5166 (vect_transform_slp_perm_load): Assert when we may not fail.
5168 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5170 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5171 message for incompatible -msdata=* and -mcall-* options.
5173 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5175 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5176 rate for post-reload scheduling.
5178 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5180 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5182 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5184 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5185 to improve monte carlo in scimark.
5187 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5189 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5190 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5191 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5192 pentium4_cost, nocona_cost): Set reassociation width to 1.
5193 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5194 width to 2 for fp operations and 1 otherwise.
5195 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5197 (atom_cost): Set reassociation width to 2.
5198 (slm_cost, generic_cost): Set fp reassociation width
5199 to 2 and 1 otherwise.
5200 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5201 (core_cost): Set fp reassociation width to 4 and vector to 2.
5202 (ix86_reassociation_width): Rewrite using cost table; special case
5203 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5204 and TARGET_AVX128_OPTIMAL.
5205 * config/i386/i386.h (processor_costs): Add
5206 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5207 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5208 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5209 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5210 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5211 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5213 2017-10-05 Nathan Sidwell <nathan@acm.org>
5215 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5217 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5219 * config/arm/arm.c (arm_test_fpu_data): New.
5220 (arm_run_selftests): Call arm_test_fpu_data.
5222 2017-10-04 Nathan Sidwell <nathan@acm.org>
5224 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5226 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5228 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5230 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5231 check type for aarch64_simd_valid_immediate.
5232 (aarch64_output_simd_mov_immediate): Update prototype.
5233 (aarch64_simd_valid_immediate): Update prototype.
5234 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5235 support for ORR-immediate.
5236 (and<mode>3): modified pattern to add support for BIC-immediate.
5237 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5238 now checks for valid immediate for BIC and ORR based on new enum
5240 (aarch64_output_simd_mov_immediate): Function now used to output
5241 BIC/ORR imm as well based on new enum argument.
5242 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5244 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5245 (aarch64_reg_or_bic_imm): Likewise.
5247 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5249 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5250 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5252 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5256 PR rtl-optimization/82396
5257 * haifa-sched.c (autopref_multipass_init): Simplify
5259 (autopref_rank_data): Simplify sort order.
5260 * sched-int.h (autopref_multipass_data_): Remove
5261 multi_mem_insn_p, min_offset and max_offset.
5263 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5265 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5267 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5269 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5272 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5274 * doc/sourcebuild.texi: Document vect_long_mult.
5276 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5278 PR tree-optimization/82413
5279 * fold-const.c (build_range_check): Use widest_int when comparing
5280 the maximum ETYPE value with HIGH.
5282 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5284 PR rtl-optimization/82396
5285 * haifa-sched.c (autopref_multipass_init): Simplify
5287 (autopref_rank_data): Simplify sort order.
5288 * sched-int.h (autopref_multipass_data_): Remove
5289 multi_mem_insn_p, min_offset and max_offset.
5291 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5293 PR tree-optimization/82381
5294 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5295 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5298 PR tree-optimization/82374
5299 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5300 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5301 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5302 current_function_decl to the new decl.
5304 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5306 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5307 helper macro for IEEE float128 hardware built-in functions.
5308 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5310 (TRUNCF128_ODD): Likewise.
5311 (ADDF128_ODD): Likewise.
5312 (SUBF128_ODD): Likewise.
5313 (MULF128_ODD): Likewise.
5314 (DIVF128_ODD): Likewise.
5315 (FMAF128_ODD): Likewise.
5316 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5317 UNSPEC_TRUNC_ROUND_TO_ODD.
5318 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5319 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5320 floating point round to odd instructions.
5321 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5322 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5323 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5324 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5325 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5326 (trunc<mode>sf2_hw): Change the truncate with round to odd
5327 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5328 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5329 to odd hardware instructions.
5330 (sub<mode>3_odd): Likewise.
5331 (mul<mode>3_odd): Likewise.
5332 (div<mode>3_odd): Likewise.
5333 (sqrt<mode>2_odd): Likewise.
5334 (fma<mode>4_odd): Likewise.
5335 (fms<mode>4_odd): Likewise.
5336 (nfma<mode>4_odd): Likewise.
5337 (nfms<mode>4_odd): Likewise.
5338 (trunc<mode>df2_odd): Change the truncate with round to odd
5339 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5341 * doc/extend.texi (PowerPC built-in functions): Update documentation
5342 for existing IEEE float128-bit built-in functions. Add built-in
5343 functions that generate the IEEE 128-bit floating point round to
5346 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5348 PR rtl-optimization/77729
5349 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5350 to (X&(C1&~C2))|C2 transformations.
5352 2017-10-03 Martin Jambor <mjambor@suse.cz>
5354 PR tree-optimization/82363
5355 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5356 mismatch, mark lacc written regardless of racc.
5358 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5360 PR tree-optimization/82381
5361 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5362 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5363 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5364 fallthrough into reassoc_stmt_dominates_stmt_p.
5367 * combine.c (combine_instructions): Don't combine in unreachable
5370 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5374 function to not use the have_cpu variable. Do not set cpu_index,
5375 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5377 (rs6000_valid_attribute_p): Remove duplicate initializations of
5378 old_optimize and func_optimize.
5379 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5380 (rs6000_activate_target_options): Make global.
5381 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5384 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5386 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5387 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5388 Set *poffset to 0 and *psize and *pmax_size to -1 if
5389 *poffset + *psize overflows in HOST_WIDE_INT.
5391 PR tree-optimization/82387
5392 PR tree-optimization/82388
5393 PR tree-optimization/82389
5394 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5395 instead of live_bytes non-NULL.
5397 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5400 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5403 2017-10-02 Richard Biener <rguenther@suse.de>
5405 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5406 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5407 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5409 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5411 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5412 requested precision matches the type's.
5413 * calls.c (alloc_max_size): Calculate the new candidate size as
5414 a widest_int and use wi::to_widest when comparing it with the
5415 current candidate size.
5416 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5417 zero rather than integer_zero_node.
5418 * match.pd: Check for a no-op conversion before using wi::add
5419 rather than after. Use tree_to_uhwi when summing small shift
5420 counts into an unsigned int.
5422 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5423 Alan Hayward <alan.hayward@arm.com>
5424 David Sherwood <david.sherwood@arm.com>
5427 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5428 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5429 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5430 POINTER_AND_FP_REGS.
5432 2017-10-02 Richard Biener <rguenther@suse.de>
5434 PR tree-optimization/82355
5435 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5436 a mapping for the enclosing loop but avoid generating one for
5438 (copy_bb_and_scalar_dependences): Remove premature codegen
5439 error on PHIs in blocks duplicated into multiple places.
5440 * graphite-scop-detection.c
5441 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5442 in the region use it as loop and nest to analyze the DR in.
5443 (try_generate_gimple_bb): Likewise.
5444 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5445 (add_loop_constraints): For blocks in a loop not in the region
5446 create a dimension with a single iteration.
5447 * sese.h (gbb_loop_at_index): Remove assert.
5449 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5451 * omp-expand.c (adjust_context_scope): New function.
5452 (expand_parallel_call): Call adjust_context_scope.
5454 2017-10-01 Jeff Law <law@redhat.com>
5456 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5457 dom_opt_dom_walker class with direct access to private members.
5458 Add comments. Call test_for_singularity.
5459 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5460 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5461 m_dummy_cond anymore.
5462 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5464 (pass_dominator:execute): Build the dummy_cond here and pass it
5465 to the dom_opt_dom_walker ctor.
5466 (test_for_singularity): New function.
5468 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5469 Maya Rashish <coypu@sdf.org>
5471 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5472 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5473 (alpha*-*-netbsd*) Use nbsd_tm_file.
5474 (arm*-*-netbsdelf*) Likewise.
5475 (i[34567]86-*-netbsdelf*) Likewise.
5476 (x86_64-*-netbsd*) Likewise.
5477 (mips*-*-netbsd*) Likewise.
5478 (powerpc-*-netbsd*) Likewise.
5479 (sh*-*-netbsd*) Likewise.
5480 (sparc-*-netbsdelf*) Likewise.
5481 (sparc64-*-netbsd*) Likewise.
5482 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5484 (vax-*-netbsdelf*) Likewise.
5485 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5486 (UINT_FAST8_TYPE) Likewise.
5487 (INT_FAST16_TYPE) Check CHAR_FAST16.
5488 (UINT_FAST16_TYPE) Likewise.
5490 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5493 * config/i386/i386.md
5494 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5495 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5496 *divmodsi4_zext_2): New define_insn_and_split.
5497 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5498 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5499 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5500 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5501 New define_insn_and_split.
5502 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5503 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5504 operands[1] having DImode when mode is SImode.
5506 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5507 always SImode for DIV and MOD in REG_EQUAL notes.
5509 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5512 * match.pd: Fix handling of NaNs in pattern.
5514 2017-09-29 Jeff Law <law@redhat.com>
5516 * sbitmap.c (bitmap_bit_in_range_p): New function.
5517 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5518 * tree-ssa-dse.c (live_bytes_read): New function.
5519 (dse_classify_store): Ignore reads of dead bytes.
5521 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5522 typos and whitespace errors.
5523 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5524 * config/s390/s390.c (s390_emit_prologue): Likewise.
5526 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5529 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5530 with a symbol for LRA.
5532 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5534 PR rtl-optimization/82338
5535 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5537 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5539 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5540 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5541 (qsort_chk): Declare.
5542 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5543 (qsort_chk): New function.
5545 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5547 PR tree-optimization/82337
5548 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5549 phi definition if the PHI result appears in an abnormal PHI.
5550 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5551 basis appears in an abnormal PHI.
5553 2017-09-29 Richard Biener <rguenther@suse.de>
5555 * graphite-isl-ast-to-gimple.c
5556 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5557 (binary_op_to_tree): Use it.
5558 (get_rename_from_scev): Likewise.
5559 (copy_loop_phi_nodes): Likewise.
5560 (copy_bb_and_scalar_dependences): Likewise.
5561 (translate_pending_phi_nodes): Likewise.
5563 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5566 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5567 for movabsq $(i32 << shift), r64.
5569 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5571 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5572 index when encoding %esp as %rsp to avoid 0x67 prefix.
5574 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5576 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5577 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5579 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5581 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5582 Extensions with more than 16 double VFP registers.
5583 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5584 to_clear_mask and all code related to it. Replace the remaining
5585 entry by a sbitmap and adapt code accordingly.
5587 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5589 * brig-builtins.def: Change pure attributes to const.
5591 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5593 * config.gcc (default_gnu_indirect_function): Default to yes for
5594 sparc*-*-linux* with glibc.
5596 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5598 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5599 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5600 when creating .init_array and .fini_array sections with priority
5603 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5606 * config/aarch64/aarch64.c
5607 (aarch64_builtin_support_vector_misalignment): Always return false
5608 when misalignment is unknown.
5610 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5612 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5613 this function to return false if the definition used by the swap
5614 instruction is artificial, or if the memory address from which the
5615 constant value is loaded is not represented by a base address held
5616 in a register or if the base address register is a frame or stack
5617 pointer. Additionally, return false if the base address of the
5618 loaded constant is a SYMBOL_REF but is not considered to be a
5620 (replace_swapped_load_constant): New function.
5621 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5622 loaded constant vector with a load of a swapped constant vector.
5624 2017-09-27 Carl Love <cel@us.ibm.com>
5626 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5627 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5628 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5631 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5633 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5634 first, always call autopref_rank_data otherwise.
5636 2017-09-27 Richard Biener <rguenther@suse.de>
5638 * graphite-scop-detection.c (find_scop_parameters): Move
5639 loop bound handling ...
5640 (gather_bbs::before_dom_children): ... here, avoiding the need
5641 to build scop_info->loop_nest.
5642 (record_loop_in_sese): Remove.
5643 * sese.h (sese_info_t::loop_nest): Remove.
5644 * sese.c (new_sese_info): Do not allocate loop_nest.
5645 (free_sese_info): Do not free loop_nest.
5647 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5650 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5651 lhs from calls if the lhs has addressable type.
5653 2017-09-27 Richard Biener <rguenther@suse.de>
5655 * graphite.h (scop::max_alias_set): New member.
5656 * graphite-scop-detection.c: Remove references to non-existing
5657 --param in comments.
5658 (build_alias_sets): Record the maximum alias set used for drs.
5659 (build_scops): Support zero as unlimited for
5660 --param graphite-max-arrays-per-scop.
5661 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5663 (build_poly_sr_1): ... here. Compute alias set based on the
5664 maximum alias set used for drs rather than
5665 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5667 2017-09-27 Richard Biener <rguenther@suse.de>
5669 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5670 --param loop-block-tile-size=0 to disable tiling.
5672 2017-09-27 Richard Biener <rguenther@suse.de>
5674 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5675 (graphite-max-nb-scop-params): Document special value zero.
5676 * domwalk.h (dom_walker::STOP): New symbolical constant.
5677 (dom_walker::dom_walker): Add optional parameter for bb to
5679 (dom_walker::~dom_walker): Declare.
5680 (dom_walker::before_dom_children): Document STOP return value.
5681 (dom_walker::m_user_bb_to_rpo): New member.
5682 (dom_walker::m_bb_to_rpo): Likewise.
5683 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5684 mapping here if not provided by the user.
5685 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5686 provided by the user.
5687 (dom_walker::STOP): Define.
5688 (dom_walker::walk): Do not compute bb to RPO mapping here.
5689 Support STOP return value from before_dom_children to stop
5691 * graphite-optimize-isl.c (optimize_isl): If the schedule
5692 is the same still generate code if -fgraphite-identity
5693 or -floop-parallelize-all are given.
5694 * graphite-scop-detection.c: Include cfganal.h.
5695 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5697 (gather_bbs::before_dom_children): Return STOP for BBs
5699 (build_scops): Compute bb to RPO mapping and pass it to
5700 the domwalk. Treat --param graphite-max-nb-scop-params=0
5701 as not limiting the number of params.
5702 * graphite.c (graphite_initialize): Remove limit on the
5703 number of basic-blocks in a function.
5704 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5705 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5706 default value of 10.
5708 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5710 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5711 Adjust code to eliminate needing to do the shift right 32-bits
5712 operation after XSCVDPSPN.
5714 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5716 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5717 ((X / Y) != 0 -> X >= Y): Likewise.
5719 2017-09-26 Carl Love <cel@us.ibm.com>
5721 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5722 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5723 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5724 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5725 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5726 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5727 definitions and overloading.
5728 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5729 statement for P9V_BUILTIN_XST_LEN_R.
5730 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5731 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5732 define_expand and define_insn for the instructions and builtins.
5733 * doc/extend.texi: Update the built-in documentation file for the new
5735 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5736 define_insn for the instructions
5738 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5741 * gcc/config/netbsd-protos.h: New file.
5742 * gcc/config/netbsd.c: New file.
5743 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5744 * gcc/config/t-netbsd: New file.
5745 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5746 (tmake_file) Add t-netbsd.
5747 (extra_objs) Add netbsd.o.
5749 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5753 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5755 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5757 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5758 sign extension from a vector register to a GPR by doing a 32-bit
5759 direct move and then an EXTSW.
5760 (extendsi<mode>2 splitter): Likewise.
5761 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5762 right or vector extract after doing XSCVDPSPN. Use
5763 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5765 (movdi_from_sf_zero_ext): Likewise.
5766 (reload_gpr_from_vsxsf): Likewise.
5767 (p8_mfvsrd_4_disf): Delete, no longer used.
5768 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5769 then needing to move the SFmode to a GPR to use the XSCVDPSP
5770 instruction instead of FRSP and XSCVDPSPN.
5771 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5772 it is adjacent to the other XSCVSPDP insns.
5773 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5774 SFmode to be in traditional Altivec registers.
5775 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5776 (vsx_xscvspdpn): Likewise.
5777 (vsx_xscvspdpn_scalar): Likewise.
5779 2017-09-26 Martin Jambor <mjambor@suse.cz>
5781 * tree-sra.c (compare_access_positions): Put integral types first,
5782 stabilize sorting of integral types, remove conditions putting
5783 non-full-precision integers last.
5784 (sort_and_splice_var_accesses): Disable scalarization if a
5785 non-integert would be represented by a non-full-precision integer.
5787 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5789 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5790 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5791 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5792 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5793 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5794 conditionals inside the function instead of around it. Call
5795 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5796 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5798 2017-09-26 Richard Biener <rguenther@suse.de>
5800 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5802 (scop_detection::build_scop_breadth): ... this. Removed.
5803 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5804 (scop_detection::harmful_stmt_in_bb): Likewise.
5805 (scop_detection::graphite_can_represent_stmt): Likewise.
5806 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5807 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5808 (scop_detection::can_represent_loop_1): ... this. Removed.
5809 (scop_detection::harmful_loop_in_region): Simplify after inlining
5810 the above and remove more quadraticness.
5811 (build_scops): Adjust.
5812 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5815 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5818 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5819 REGNO (base) == SP_REG if base is a REG.
5822 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5823 if it is different SSA_NAME.
5824 (optimize_range_tests_cmp_bitwise): New function.
5825 (optimize_range_tests): Call it.
5827 2017-09-26 Richard Biener <rguenther@suse.de>
5829 PR tree-optimization/82321
5830 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5831 for the def being inside the loop.
5833 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5835 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5837 * config/s390/s390-builtins.def: Fix constraint on op4.
5839 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5841 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5842 independent expanders.
5843 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5844 ("vec_ordered", "vec_unordered"): New expanders.
5846 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5848 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5851 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5853 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5854 vec_unpacks_lo_v16qi.
5855 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5857 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5859 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5860 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5861 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5863 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5865 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5867 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5869 ("*vec_slb<mode>"): New insn pattern.
5870 ("vec_shr_<mode>"): New expander.
5871 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5872 and force the shift count operand to V16QImode.
5873 ("vec_srb<mode>"): Set shift count mode to V16QI.
5875 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5877 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5878 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5879 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5881 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5884 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5886 2017-09-26 Richard Biener <rguenther@suse.de>
5888 PR tree-optimization/82320
5889 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5892 2017-09-25 Jeff Law <law@redhat.com>
5894 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5895 prototype for new argument.
5896 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5897 mostly extracted from rs6000_emit_allocate_stack.
5898 (rs6000_emit_probe_stack_range_stack_clash): New function.
5899 (rs6000_emit_allocate_stack): Call
5900 rs6000_emit_probe_stack_range_stack_clash as needed.
5901 (rs6000_emit_probe_stack_range): Add additional argument
5902 to call to gen_probe_stack_range{si,di}.
5903 (output_probe_stack_range): New.
5904 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5905 (output_probe_stack_range_stack_clash): New.
5906 (rs6000_emit_prologue): Emit notes into dump file as requested.
5907 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5908 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5909 Add additional operand and pass it to output_probe_stack_range.
5911 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5913 PR tree-optimization/82163
5914 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5915 (checking_verify_loop_closed_ssa): New parameter.
5916 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5917 (check_loop_closed_ssa_stmt): Delete.
5918 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5919 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5920 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5923 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5925 * brig-builtins.def: Treat HSAIL barrier builtins as
5926 setjmp/longjump style functions.
5928 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5930 * target.def (constant_alignment): New hook.
5931 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5932 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5933 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5934 * doc/tm.texi: Regenerate.
5935 * targhooks.h (default_constant_alignment): Declare.
5936 (constant_alignment_word_strings): Likewise.
5937 * targhooks.c (default_constant_alignment): New function.
5938 (constant_alignment_word_strings): Likewise.
5939 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5940 instead of CONSTANT_ALIGNMENT.
5941 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5942 (force_const_mem): Likewise.
5943 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5944 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5945 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5946 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5947 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5949 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5950 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5951 constant_alignment_word_strings.
5952 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5953 (CONSTANT_ALIGNMENT): Likewise.
5954 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5955 (arm_constant_alignment): New function.
5956 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5957 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5958 constant_alignment_word_strings.
5959 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5960 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5961 constant_alignment_word_strings.
5962 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5963 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5964 (cris_constant_alignment): New function.
5965 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5966 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5967 (epiphany_constant_alignment): New function.
5968 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5969 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5970 constant_alignment_word_strings.
5971 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5972 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5973 constant_alignment_word_strings.
5974 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5975 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5976 constant_alignment_word_strings.
5977 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5978 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5979 * config/i386/i386.c (ix86_constant_alignment): Make static.
5980 Use the same interface as the target hook.
5981 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5982 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5983 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5984 constant_alignment_word_strings.
5985 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5986 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5987 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5988 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5989 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5990 constant_alignment_word_strings.
5991 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5992 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5993 constant_alignment_word_strings.
5994 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5995 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5996 constant_alignment_word_strings.
5997 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5998 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6000 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6001 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6002 * config/mips/mips.c (mips_constant_alignment): New function.
6003 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6004 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6005 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6006 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6007 (mmix_constant_alignment): Make static. Use the same interface
6009 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6010 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6011 constant_alignment_word_strings.
6012 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6013 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6014 constant_alignment_word_strings.
6015 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6016 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6017 constant_alignment_word_strings.
6018 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6019 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6020 (rs6000_constant_alignment): New function.
6021 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6022 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6023 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6024 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6025 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6026 (rs6000_constant_alignment): New function.
6027 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6028 * config/s390/s390.c (s390_constant_alignment): New function.
6029 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6030 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6031 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6032 constant_alignment_word_strings.
6033 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6034 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6035 (sparc_constant_alignment): New function.
6036 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6037 * config/spu/spu.c (spu_constant_alignment): New function.
6038 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6039 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6040 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6041 constant_alignment_word_strings.
6042 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6043 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6044 constant_alignment_word_strings.
6045 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6046 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6047 constant_alignment_word_strings.
6048 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6049 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6050 (visium_constant_alignment): New function.
6051 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6052 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6053 (xtensa_constant_alignment): New function.
6054 * system.h (CONSTANT_ALIGNMENT): Poison.
6056 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6059 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6060 (rs6000_builtin_valid_without_lhs): New helper function.
6061 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6062 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6064 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6066 * target.h (vec_perm_indices): Use unsigned short rather than
6068 (auto_vec_perm_indices): Likewise.
6069 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6070 Use unsigned int rather than unsigned char.
6071 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6073 2017-09-25 Richard Biener <rguenther@suse.de>
6075 * cfgloop.h (sort_sibling_loops): Declare.
6076 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6077 (sort_sibling_loops): New function sorting the sibling loop list
6079 * graphite.c (graphite_transform_loops): Sort sibling loops.
6081 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6083 * target.def (vec_perm_const_ok): Change sel parameter to
6085 * optabs-query.c (can_vec_perm_p): Update accordingly.
6086 * doc/tm.texi: Regenerate.
6087 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6088 auto_vec_perm_indices and remove separate nelt field.
6089 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6090 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6091 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6092 (aarch64_expand_vec_perm_const): Update accordingly.
6093 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6094 to vec_perm_indices.
6095 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6096 auto_vec_perm_indices and remove separate nelt field.
6097 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6098 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6099 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6101 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6102 to vec_perm_indices.
6103 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6104 sel to vec_perm_indices.
6105 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6106 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6107 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6109 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6112 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6115 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6116 on the FUNCTION_DECL function context if it has a DIE that is a
6119 2017-09-25 Richard Biener <rguenther@suse.de>
6121 PR tree-optimization/82285
6122 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6125 2017-09-25 Tom de Vries <tom@codesourcery.com>
6129 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6132 2017-09-25 Richard Biener <rguenther@suse.de>
6134 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6135 ISL errors other than isl_error_quota happen. Dump if the
6136 schedule is the same.
6137 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6138 errors instead of aborting inside ISL.
6140 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6143 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6144 of libgcc_eh for m64.
6145 * config/i386/darwin64.h: Likewise.
6147 2017-09-25 Richard Biener <rguenther@suse.de>
6150 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6151 attribute for incomplete types nor twice for complete ones.
6153 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6156 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6157 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6160 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6163 * config/i386/i386.opt (mprefer-avx256): Use
6164 ix86_target_flags variable.
6165 * config/i386/i386.c (ix86_target_string): Move
6166 -mprefer-avx256 to flag2_opts.
6168 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6171 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6172 and x != -1 | y != -1 into (x & y) != -1.
6174 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6176 * config.gcc: Add new case statement to set
6177 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6178 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6179 s390x-*-linux* case statements. Added aarch64 to the list of
6180 supported architectures.
6182 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6184 PR tree-optimization/82289
6185 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6186 STMT_VINFO_RELEVANT_P.
6188 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6189 Alan Hayward <alan.hayward@arm.com>
6190 David Sherwood <david.sherwood@arm.com>
6192 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6193 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6194 (extract_range_from_binary_expr_1): Don't call
6195 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6197 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6198 Alan Hayward <alan.hayward@arm.com>
6199 David Sherwood <david.sherwood@arm.com>
6201 * target.def (preferred_vector_alignment): New hook.
6202 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6204 * doc/tm.texi: Regenerate.
6205 * targhooks.h (default_preferred_vector_alignment): Declare.
6206 * targhooks.c (default_preferred_vector_alignment): New function.
6207 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6209 (DR_TARGET_ALIGNMENT): New macro.
6210 (aligned_access_p): Update commentary.
6211 (vect_known_alignment_in_bytes): New function.
6212 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6214 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6215 Calculate the misalignment based on the target alignment rather than
6217 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6218 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6219 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6220 the target alignment, rather than masking the element misalignment
6221 with the number of elements in a vector. Also use the target
6222 alignment when calculating the maximum number of peels.
6223 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6224 instead of TYPE_ALIGN_UNIT.
6225 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6226 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6227 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6228 (vect_create_data_ref_ptr): Likewise.
6229 (vect_setup_realignment): Realign by ANDing with
6230 -DR_TARGET_MISALIGNMENT.
6231 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6232 the number of peels based on DR_TARGET_ALIGNMENT.
6233 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6234 with the guaranteed alignment boundary when deciding whether
6236 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6237 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6238 (ensure_base_align): Remove stmt_info parameter. Get the
6239 target base alignment from DR_TARGET_ALIGNMENT.
6240 (vectorizable_store): Update call accordingly. Interpret
6241 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6243 (vectorizable_load): Likewise.
6245 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6246 Alan Hayward <alan.hayward@arm.com>
6247 David Sherwood <david.sherwood@arm.com>
6249 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6250 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6251 (vect_enhance_data_refs_alignment): Likewise.
6253 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6255 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6256 error. Only quit immediately if parsing is complete.
6257 (BEGIN): Initialize fatal_err and parse_done.
6258 (begin fpu, end fpu): Check number of arguments.
6259 (begin arch, end arch): Likewise.
6260 (begin cpu, end cpu): Likewise.
6261 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6262 (optalias): Likewise.
6264 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6266 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6267 * config/arm/arm-isa.h: Delete. Move definitions to ...
6268 * arm-cpus.in: ... here. Use new feature and fgroup values.
6269 * config/arm/arm.c (arm_option_override): Use lower case for feature
6271 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6272 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6273 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6275 (print_isa_bits_for): New function.
6276 (gen_isa): New function.
6277 (gen_comm_data): Use print_isa_bits_for.
6278 (define feature): New keyword.
6279 (define fgroup): New keyword.
6280 * config/arm/t-arm (TM_H): Remove.
6281 (GTM_H): Add arm-isa.h.
6282 (arm-isa.h): Add rule to generate file.
6283 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6284 case for feature bit names.
6286 2017-09-22 Richard Biener <rguenther@suse.de>
6288 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6290 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6291 print of no dependency loops ...
6292 * graphite.c (graphite_transform_loops): ... here.
6293 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6295 (same_close_phi_node, remove_duplicate_close_phi,
6296 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6297 (canonicalize_loop_closed_ssa): ... here and simplify.
6298 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6299 (optimize_isl): Use dump_printf_loc to tell when we stopped
6300 optimizing because of an ISL timeout.
6302 2017-09-22 Richard Biener <rguenther@suse.de>
6304 PR tree-optimization/82291
6305 * tree-if-conv.c (predicate_mem_writes): Make sure to
6306 remove writes in blocks predicated with false.
6308 2017-09-22 Richard Biener <rguenther@suse.de>
6310 * sese.c: Include cfganal.h.
6311 (if_region_set_false_region): Remove.
6312 (create_if_region_on_edge): Likewise.
6313 (move_sese_in_condition): Re-implement without destroying
6316 2017-09-22 Richard Biener <rguenther@suse.de>
6318 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6319 Verify both BBs contain loop PHI nodes before dispatching to
6321 (graphite_regenerate_ast_isl): Do not recompute dominators,
6322 do not verify three times. Restructure for clarity.
6323 * graphite-scop-detection.c (same_close_phi_node,
6324 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6325 defined_in_loop_p, canonicalize_loop_closed_ssa,
6326 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6327 checking and SSA rewrite, move to ...
6328 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6329 (graphite_initialize): Do not pass in ctx, do not reset the
6330 SCEV cache, compute only dominators.
6331 (graphite_transform_loops): Allocate ISL ctx after
6332 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6333 Maintain post-dominators only around build_scops.
6334 * sese.c (if_region_set_false_region): Make static. Free
6335 and recompute dominators.
6336 (move_sese_in_condition): Assert we don't get called with
6337 post-dominators computed.
6338 * sese.h (if_region_set_false_region): Remove.
6340 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6342 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6343 mode attribute for TARGET_AVX512VL.
6345 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6347 * config/i386/i386.opt (mprefer-avx256): New option.
6348 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6350 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6351 for TARGET_PREFER_AVX256.
6352 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6354 2017-09-21 Jeff Law <law@redhat.com>
6356 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6357 Fix dump output if the only stack space is for pushed registers.
6359 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6361 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6364 2017-09-21 Martin Sebor <msebor@redhat.com>
6367 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6368 code (in C++) or code that triggers warnings.
6370 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6372 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6374 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6376 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6377 * sched-rgn.c: Ditto.
6378 * sel-sched-ir.c: Ditto.
6380 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6382 * toplev.h (set_random_seed): Adjust return type.
6383 * toplev.c (init_local_tick): Move eager initialization of random_seed
6384 to get_random_seed. Adjust comment.
6385 (init_random_seed): Inline to get_random_seed, delete.
6386 (get_random_seed): Initialize random_seed lazily.
6387 (set_random_seed): Do not return previous value.
6388 (print_switch_value): Do not call get_random_seed.
6390 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6392 * cgraph.c (delete_function_version): New, broken out from...
6393 (cgraph_node::delete_function_version): ...here. Rename to
6394 cgraph_node::delete_function_version_by_decl. Update all uses.
6395 (cgraph_node::remove): Call delete_function_version.
6397 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6400 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6401 VAR_DECLs to which addressable non-volatile parameters are mapped
6402 and for id->retvar after the return value assignment. Clear
6403 id->retval and id->retbnd after inlining.
6405 2017-09-21 Richard Biener <rguenther@suse.de>
6407 PR tree-optimization/82276
6408 PR tree-optimization/82244
6409 * tree-vrp.c (build_assert_expr_for): Set
6410 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6412 (remove_range_assertions): Revert earlier change.
6414 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6417 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6419 2017-09-21 Richard Biener <rguenther@suse.de>
6421 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6422 Restore valid IL after code generation errors.
6423 * graphite.c (graphite_transform_loops): Diagnose code
6424 generation issues as MSG_MISSED_OPTIMIZATION and continue
6425 with processing SCOPs.
6427 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6428 Alan Hayward <alan.hayward@arm.com>
6429 David Sherwood <david.sherwood@arm.com>
6431 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6432 rather than choosing between plus_constant and gen_rtx_<CODE>.
6433 * expr.c (emit_push_insn): Likewise.
6434 (expand_expr_real_2): Likewise.
6436 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6437 Alan Hayward <alan.hayward@arm.com>
6438 David Sherwood <david.sherwood@arm.com>
6440 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6442 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6443 Alan Hayward <alan.hayward@arm.com>
6444 David Sherwood <david.sherwood@arm.com>
6446 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6447 calling tree_to_uhwi.
6449 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6450 Alan Hayward <alan.hayward@arm.com>
6451 David Sherwood <david.sherwood@arm.com>
6453 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6454 INTEGER_CST rather than a negative test for ADDR_EXPR.
6456 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6457 Alan Hayward <alan.hayward@arm.com>
6458 David Sherwood <david.sherwood@arm.com>
6460 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6461 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6463 2017-09-21 Richard Biener <rguenther@suse.de>
6465 PR tree-optimization/71351
6466 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6467 graphite_create_new_loop_guard): Remove, fold remaining parts
6469 (translate_isl_ast_node_for): ... here and simplify.
6471 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6474 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6475 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6476 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6477 alternative always use QI mode, for -Os imov (=R,R) alternative
6478 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6481 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6482 Jeff Law <law@redhat.com>
6484 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6485 (allocate_stack_space): New function, partially extracted from
6487 (s390_emit_prologue): Track offset to most recent stack probe.
6488 Code to allocate space moved into allocate_stack_space.
6489 Dump actions when no stack is allocated.
6490 (s390_prologue_plus_offset): New function.
6491 (s390_emit_stack_probe): Likewise.
6493 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6495 * common.opt (Wa, Wl, Wp, g, gz=): Add
6497 (gno-column-info): Remove.
6498 (gcolumn-info): Drop RejectNegative.
6500 (gno-record-gcc-switches): Remove.
6501 (grecord-gcc-switches): Drop RejectNegative.
6502 (gno-split-dwarf): Remove.
6503 (gsplit-dwarf): Drop RejectNegative.
6504 (gno-strict-dwarf): Remove.
6505 (gstrict-dwarf): Drop RejectNegative.
6506 * config/darwin.opt (gfull, gused): Add RejectNegative.
6507 * dwarf2out.c (gen_producer_string): Drop
6508 gno-record-gcc-switches handler.
6509 * optc-gen.awk: Add g to prefixes with negative forms.
6510 * opts-common.c (remapping_prefix_p): New.
6511 (find_opt): Check it.
6512 (generate_canonical_option): Test g prefix.
6513 (option_map): Add -gno- mapping.
6514 (add_misspelling_candidates): Check remapping_prefix_p.
6516 2017-09-20 Jeff Law <law@redhat.com>
6518 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6519 thinko in stack clash protection support.
6521 * explow.c (compute_stack_clash_protection_loop_data): Use
6522 CONST_INT_P instead of explicit test. Verify object is a
6523 CONST_INT_P before looking at INTVAL.
6524 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6525 instead of explicit test.
6527 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6530 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6531 address instead of to r1 and r11.
6533 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6535 * config.gcc: Support "knm".
6536 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6537 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6539 * config/i386/i386.c (m_KNM): Define.
6540 (processor_target_table): Add "knm".
6542 (ix86_option_override_internal): Add "knm".
6543 (ix86_issue_rate): Add PROCESSOR_KNM.
6544 (ix86_adjust_cost): Ditto.
6545 (ia32_multipass_dfa_lookahead): Ditto.
6546 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6547 (fold_builtin_cpu): Add M_INTEL_KNM.
6548 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6549 (processor_type): Add PROCESSOR_KNM.
6550 * config/i386/x86-tune.def: Add m_KNM.
6551 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6553 2017-09-20 Richard Biener <rguenther@suse.de>
6555 PR tree-optimization/80213
6556 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6557 are allowed in empty BBs as well.
6558 (canonicalize_loop_closed_ssa): Also look for other complex
6560 (scop_detection::get_sese): Include the loop-closed PHI block
6562 (scop_detection::merge_sese): Remove code adding extra blocks.
6563 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6564 (build_scops): Assert the final returned scop is invalid.
6566 2017-09-20 Richard Biener <rguenther@suse.de>
6568 PR tree-optimization/82264
6569 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6571 (vn_phi_lookup): Likewise.
6572 (vn_phi_insert): Likewise.
6574 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6576 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6577 that fits into uhwi or shwi, add DW_AT_const_value regardless
6578 of early_dwarf without going through RTL, using add_AT_unsigned
6581 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6582 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6583 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6584 (DEBUG_MACRO_SECTION): Likewise.
6585 (DEBUG_LTO_MACRO_SECTION): Likewise.
6586 (DEBUG_STR_DWO_SECTION): Likewise.
6587 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6588 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6589 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6590 (DEBUG_LTO_LINE_STR_SECTION): Define.
6591 (init_sections_and_labels): Initialize debug_line_str_section
6592 variable. Initialize debug_loc_section for -gdwarf-5 to
6593 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6595 2017-09-20 Richard Biener <rguenther@suse.de>
6597 * graphite-sese-to-poly.c (extract_affine): Properly handle
6598 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6600 2017-09-20 Richard Biener <rguenther@suse.de>
6602 PR tree-optimization/81373
6603 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6604 Force SESE live-out defs to be handled even if they are
6607 2017-09-19 Jeff Law <law@redhat.com>
6609 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6610 nothing for stack adjustments with REG_STACK_CHECK.
6611 * sched-deps.c (parse_add_or_inc): Reject insns with
6612 REG_STACK_CHECK from dependency breaking.
6613 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6614 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6615 * reg-notes.def (STACK_CHECK): New note.
6617 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6618 (ix86_expand_prologue): Dump stack clash info as needed.
6619 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6621 * function.c (dump_stack_clash_frame_info): New function.
6622 * function.h (dump_stack_clash_frame_info): Prototype.
6623 (enum stack_clash_probes): New enum.
6625 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6626 flag_stack_clash_protection.
6627 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6628 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6629 (arm_frame_pointer_required): Likewise.
6630 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6631 (ia64_expand_prologue): Likewise.
6632 * config/mips/mips.c (mips_expand_prologue): Likewise.
6633 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6634 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6635 (sparc_flat_expand_prologue): Likewise.
6636 * config/spu/spu.c (spu_expand_prologue): Likewise.
6638 * explow.c: Include "params.h".
6639 (anti_adjust_stack_and_probe_stack_clash): New function.
6640 (get_stack_check_protect): Likewise.
6641 (compute_stack_clash_protection_loop_data): Likewise.
6642 (emit_stack_clash_protection_loop_start): Likewise.
6643 (emit_stack_clash_protection_loop_end): Likewise.
6644 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6645 Use anti_adjust_stack_and_probe_stack_clash.
6646 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6647 (emit_stack_clash_protection_loop_start): Likewise.
6648 (emit_stack_clash_protection_loop_end): Likewise.
6649 * rtl.h (get_stack_check_protect): Prototype.
6650 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6651 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6652 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6654 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6656 * doc/tm.texi: Rebuilt.
6657 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6658 get_stack_check_protect.
6659 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6660 * config/arm/arm.c (arm_expand_prologue): Likewise.
6661 (arm_frame_pointer_required): Likewise.
6662 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6663 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6664 * config/mips/mips.c (mips_expand_prologue): Likewise.
6665 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6666 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6667 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6668 (sparc_flat_expand_prologue): Likewise.
6670 * common.opt (-fstack-clash-protection): New option.
6671 * flag-types.h (enum stack_check_type): Note difference between
6672 -fstack-check= and -fstack-clash-protection.
6673 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6674 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6675 * toplev.c (process_options): Issue warnings/errors for cases
6676 not handled with -fstack-clash-protection.
6677 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6678 (-fstack-check): Note additional problem with -fstack-check=generic.
6679 Note that -fstack-check is primarily for Ada and refer users
6680 to -fstack-clash-protection for stack-clash-protection.
6681 Document new params for stack clash protection.
6683 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6685 * config/i386/i386.c (ix86_split_long_move): Do not handle
6686 address used for LEA in a special way.
6688 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6690 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6692 2017-09-19 Martin Sebor <msebor@redhat.com>
6695 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6696 of incompatible types.
6698 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6700 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6701 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6702 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6703 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6705 2017-09-19 Richard Biener <rguenther@suse.de>
6707 PR tree-optimization/82244
6708 * tree-vrp.c (remove_range_assertions): Do not propagate
6709 a constant to abnormals but replace the assert with a copy.
6711 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6713 PR rtl-optimization/57878
6714 PR rtl-optimization/68988
6715 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6716 avoidance test involving non_reload_pseudos. Move frequency test
6717 below the general fragmentation avoidance test.
6719 2017-09-19 Richard Biener <rguenther@suse.de>
6721 PR tree-optimization/69728
6722 * graphite-sese-to-poly.c (schedule_error): New global.
6723 (add_loop_schedule): Handle empty domain by failing the
6725 (build_original_schedule): Handle schedule_error.
6727 2017-09-19 Richard Biener <rguenther@suse.de>
6729 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6730 Do not iterate to sibling loops but only to siblings of inner
6733 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6736 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6737 that the registers are different.
6739 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6741 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6742 to processor_model and "amdfam17h" to arch_names_table.
6743 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6745 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6748 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6750 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6751 Alan Hayward <alan.hayward@arm.com>
6752 David Sherwood <david.sherwood@arm.com>
6754 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6756 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6758 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6759 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6760 vect_schedule_slp_instance.
6761 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6762 Update call to vect_slp_analyze_node_operations. Simplify return
6764 (vect_slp_analyze_bb_1): Update call accordingly.
6765 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6766 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6767 (vect_schedule_slp): Update call accordingly.
6769 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6770 Alan Hayward <alan.hayward@arm.com>
6771 David Sherwood <david.sherwood@arm.com>
6773 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6774 with types that aren't in fact scalar.
6776 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6778 * tree-vect-slp.c (vect_record_max_nunits): New function,
6780 (vect_build_slp_tree_1): ...here.
6781 (vect_build_slp_tree_2): Call it for phis too.
6783 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6785 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6786 to vect_get_vec_def_for_operand when getting the mask operand.
6788 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6789 Alan Hayward <alan.hayward@arm.com>
6790 David Sherwood <david.sherwood@arm.com>
6792 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6795 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6796 Alan Hayward <alan.hayward@arm.com>
6797 David Sherwood <david.sherwood@arm.com>
6799 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6800 calculation for vector booleans.
6802 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6803 Alan Hayward <alan.hayward@arm.com>
6804 David Sherwood <david.sherwood@arm.com>
6806 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6808 (vect_transform_stmt): ...here.
6809 (vect_analyze_stmt): Use it instead of calling
6810 vectorizable_live_operation directly.
6812 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6814 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6815 non-SIMT targets in acc vector loops.
6817 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6819 * configure.ac: Add arc and check if assembler supports gdwarf2.
6820 * configure: Regenerate.
6822 2017-09-18 Richard Biener <rguenther@suse.de>
6824 PR tree-optimization/82220
6825 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6826 epilogue niters from the min_profitable_iters compute.
6828 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6831 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6832 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6833 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6835 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6838 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6839 switching to a new text section.
6841 2017-09-18 Richard Biener <rguenther@suse.de>
6843 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6845 (build_alias_set): Reject aliases with no access function.
6847 2017-09-18 Richard Biener <rguenther@suse.de>
6849 PR tree-optimization/79622
6850 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6852 (build_cross_bb_scalars_use): Likewise.
6854 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6856 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6858 2017-09-18 Alan Modra <amodra@gmail.com>
6861 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6862 stack_pointer_rtx for count 0. Update comments. Break up
6863 large rtl expression.
6865 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6867 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6868 Increase to 20 bytes.
6869 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6870 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6871 or avx version of the stub.
6873 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6876 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6877 compute the minimum stack alignment. Also update preferred stack
6878 boundary for leaf functions.
6880 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6882 PR tree-optimization/82228
6883 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6886 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6888 * common/config/nds32/nds32-common.c
6889 (nds32_option_optimization_table): Refine formatting.
6890 (nds32_option_optimization_table): Use -fsched-pressure and
6891 -fomit-frame-pointer for specific optimization level.
6893 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6895 * config/nds32/nds32.c: Refine formatting and comments.
6896 * config/nds32/nds32.h: Likewise.
6897 * config/nds32/nds32.md: Likewise.
6898 * config/nds32/nds32-cost.c: Likewise.
6899 * config/nds32/nds32-isr.c: Likewise.
6900 * config/nds32/nds32-md-auxiliary.c: Likewise.
6901 * config/nds32/nds32-multiple.md: Likewise.
6902 * config/nds32/nds32-predicates.c: Likewise.
6904 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6905 Jakub Jelinek <jakub@redhat.com>
6907 Add support for -std=c++2a.
6908 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6910 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6912 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6915 * doc/extend.texi (Common Function Attributes): Add
6916 references to ARM, AArch64, and S/390 specific attributes.
6917 (Function Specific Option Pragmas): Add AArch64 and S/390
6918 to list of back ends that support the target pragma.
6920 2017-09-15 Nathan Sidwell <nathan@acm.org>
6922 * doc/standards.texi: Fix C++17 description. Update URLs for
6925 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6927 * common.opt (Wcast-align=strict): New warning option.
6928 * doc/invoke.texi: Document -Wcast-align=strict.
6930 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6932 * cgraph.h (cgraph_thunk_info): Add comments.
6933 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6934 assert for VIRTUAL_* arguments stricter.
6936 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6938 PR tree-optimization/71026
6939 * match.pd: Move RDIV patterns from fold-const.c
6940 * fold-const.c (distribute_real_division): Removed.
6941 (fold_binary_loc): Remove calls to distribute_real_divison.
6943 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6945 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6946 c++1z and gnu++1z as deprecated. Change other references to
6947 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6948 Change -Wc++1z-compat to -Wc++17-compat.
6949 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6950 * dwarf2out.c (highest_c_language): Handle C++17.
6951 (gen_compile_unit_die): Likewise.
6953 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6955 PR rtl-optimization/82192
6956 * combine.c (make_extraction): Don't look through non-paradoxical
6957 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6960 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6961 Alan Hayward <alan.hayward@arm.com>
6962 David Sherwood <david.sherwood@arm.com>
6964 * target.def (function_arg_offset): New hook.
6965 * targhooks.h (default_function_arg_offset): Declare.
6966 * targhooks.c (default_function_arg_offset): New function.
6967 * function.c (locate_and_pad_parm): Use
6968 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6969 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6970 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6971 * doc/tm.texi: Regenerate.
6972 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6973 * config/spu/spu.c (spu_function_arg_offset): New function.
6974 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6975 * system.h (FUNCTION_ARG_OFFSET): Poison.
6977 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6978 Alan Hayard <alan.hayward@arm.com>
6979 David Sherwood <david.sherwood@arm.com>
6981 * target.def (truly_noop_truncation): New hook.
6982 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6983 than TRULY_NOOP_TRUNCATION.
6984 * hooks.h (hook_bool_uint_uint_true): Declare.
6985 * hooks.c (hook_bool_uint_uint_true): New function.
6986 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6987 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6988 * doc/tm.texi: Regenerate.
6989 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6990 rather than TRULY_NOOP_TRUNCATION in comments.
6991 (simplify_comparison): Likewise.
6992 (record_truncated_value): Likewise.
6993 * expmed.c (extract_bit_field_1): Likewise.
6994 (extract_split_bit_field): Likewise.
6995 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6996 instead of TRULY_NOOP_TRUNCATION.
6997 * function.c (assign_parm_setup_block): Likewise.
6998 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6999 * rtlhooks.c: Include target.h.
7000 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7001 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7002 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7003 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7004 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7005 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7006 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7007 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7008 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7009 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7010 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7011 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7012 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7013 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7014 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7015 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7016 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7017 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7018 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7019 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7020 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7021 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7022 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7023 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7024 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7025 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7026 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7027 rather than TRULY_NOOP_TRUNCATION in comments.
7028 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7029 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7030 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7031 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7032 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7033 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7034 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7035 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7036 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7037 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7038 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7039 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7040 rather than TRULY_NOOP_TRUNCATION in comments.
7041 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7042 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7043 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7044 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7045 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7046 TRULY_NOOP_TRUNCATION condition.
7047 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7048 (TRULY_NOOP_TRUNCATION): Delete.
7049 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7050 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7051 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7052 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7053 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7054 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7055 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7056 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7057 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7058 rather than TRULY_NOOP_TRUNCATION in comments.
7059 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7060 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7061 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7062 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7063 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7064 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7066 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7069 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7070 (*cmp_ior): Likewise.
7071 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7072 (*ior_scc_scc_cmp): Likewise.
7073 (*and_scc_scc): Likewise.
7074 (*and_scc_scc_cmp): Likewise.
7076 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7077 Alan Hayard <alan.hayward@arm.com>
7078 David Sherwood <david.sherwood@arm.com>
7080 * target.def (can_change_mode_class): New hook.
7081 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7082 (hard_regno_nregs): Likewise.
7083 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7084 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7085 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7086 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7087 (LOAD_EXTEND_OP): Update accordingly.
7088 * doc/tm.texi: Regenerate.
7089 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7090 CANNOT_CHANGE_MODE_CLASS.
7091 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7092 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7093 * combine.c (simplify_set): Update accordingly.
7094 * emit-rtl.c (validate_subreg): Likewise.
7095 * recog.c (general_operand): Likewise.
7096 * regcprop.c (mode_change_ok): Likewise.
7097 * reload1.c (choose_reload_regs): Likewise.
7098 (inherit_piecemeal_p): Likewise.
7099 * rtlanal.c (simplify_subreg_regno): Likewise.
7100 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7101 instead of CANNOT_CHANGE_MODE_CLASS.
7102 (reload_cse_simplify_operands): Likewise.
7103 * reload.c (push_reload): Use targetm.can_change_mode_class
7104 instead of CANNOT_CHANGE_MODE_CLASS.
7105 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7106 REG_CANNOT_CHANGE_MODE_P.
7107 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7108 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7109 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7110 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7111 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7112 (arm_can_change_mode_class): New function.
7113 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7114 than CANNOT_CHANGE_MODE_CLASS in comments.
7115 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7116 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7117 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7118 (ix86_can_change_mode_class): ...this new function, inverting the
7119 sense of the return value.
7120 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7121 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7122 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7123 (ia64_can_change_mode_class): New function.
7124 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7125 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7126 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7127 (m32c_can_change_mode_class): ...this new function, inverting the
7128 sense of the return value.
7129 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7130 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7131 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7132 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7133 (mips_can_change_mode_class): ...this new function, inverting the
7134 sense of the return value.
7135 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7136 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7137 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7138 (msp430_can_change_mode_class): New function.
7139 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7140 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7141 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7142 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7143 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7144 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7145 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7146 (pa_cannot_change_mode_class): Replace with...
7147 (pa_can_change_mode_class): ...this new function, inverting the
7148 sense of the return value.
7149 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7150 than CANNOT_CHANGE_MODE_CLASS in comments.
7151 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7152 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7153 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7154 (pdp11_cannot_change_mode_class): Replace with...
7155 (pdp11_can_change_mode_class): ...this new function, inverting the
7156 sense of the return value.
7157 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7158 * config/powerpcspe/powerpcspe-protos.h
7159 (rs6000_cannot_change_mode_class_ptr): Delete.
7160 * config/powerpcspe/powerpcspe.c
7161 (rs6000_cannot_change_mode_class_ptr): Delete.
7162 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7163 (rs6000_option_override_internal): Assign to
7164 targetm.can_change_mode_class instead of
7165 rs6000_cannot_change_mode_class_ptr.
7166 (rs6000_cannot_change_mode_class): Replace with...
7167 (rs6000_can_change_mode_class): ...this new function, inverting the
7168 sense of the return value.
7169 (rs6000_debug_cannot_change_mode_class): Replace with...
7170 (rs6000_debug_can_change_mode_class): ...this new function.
7171 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7172 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7173 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7174 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7175 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7177 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7178 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7179 (rs6000_option_override_internal): Assign to
7180 targetm.can_change_mode_class instead of
7181 rs6000_cannot_change_mode_class_ptr.
7182 (rs6000_cannot_change_mode_class): Replace with...
7183 (rs6000_can_change_mode_class): ...this new function, inverting the
7184 sense of the return value.
7185 (rs6000_debug_cannot_change_mode_class): Replace with...
7186 (rs6000_debug_can_change_mode_class): ...this new function.
7187 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7188 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7189 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7190 (s390_can_change_mode_class): ...this new function, inverting the
7191 sense of the return value.
7192 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7193 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7194 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7195 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7196 (sh_cannot_change_mode_class): Replace with...
7197 (sh_can_change_mode_class): ...this new function, inverting the
7198 sense of the return value.
7199 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7200 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7201 (sparc_can_change_mode_class): New function.
7202 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7203 * config/spu/spu.c (spu_can_change_mode_class): New function.
7204 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7205 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7206 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7207 (visium_can_change_mode_class): New function.
7208 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7210 2017-09-15 Richard Biener <rguenther@suse.de>
7212 PR tree-optimization/82217
7213 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7214 but not undefined case.
7216 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7219 * postreload.c (reload_cse_simplify_operands): Skip
7220 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7222 2017-09-15 Richard Biener <rguenther@suse.de>
7224 PR tree-optimization/68823
7225 * graphite-scop-detection.c (build_alias_set): If we have a
7226 possible dependence check whether we can handle them by just
7227 looking at the DRs DR_ACCESS_FNs.
7228 (build_scops): If build_alias_set fails, fail the SCOP.
7230 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7232 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7233 to support float128 built-in functions that require the ISA 3.0
7235 (BU_FLOAT128_3_HW): Likewise.
7236 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7238 (FMAF128): Likewise.
7240 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7241 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7242 floating point instructions.
7243 (rs6000_invalid_builtin): Likewise.
7244 (rs6000_builtin_mask_names): Likewise.
7245 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7246 (RS6000_BTM_FLOAT128_HW): Likewise.
7247 (RS6000_BTM_COMMON): Likewise.
7248 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7250 * doc/extend.texi (RS/6000 built-in functions): Document the
7251 IEEE 128-bit floating point square root and fused multiply-add
7254 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7256 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7257 reg (r2) isn't in the set of registers defined in the prologue.
7259 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7260 Alan Hayward <alan.hayward@arm.com>
7261 David Sherwood <david.sherwood@arm.com>
7263 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7264 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7265 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7266 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7267 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7269 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7270 max_vectorization_factor.
7271 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7273 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7274 Alan Hayward <alan.hayward@arm.com>
7275 David Sherwood <david.sherwood@arm.com>
7277 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7278 (vect_worthwhile_without_simd_p): Declare.
7279 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7280 (vectorizable_reduction): Use it.
7281 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7282 (vectorizable_operation): Likewise.
7284 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7285 Alan Hayward <alan.hayward@arm.com>
7286 David Sherwood <david.sherwood@arm.com>
7288 * tree-vectorizer.h (vect_get_num_copies): New function.
7289 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7290 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7291 (vectorizable_induction): Likewise.
7292 (vectorizable_live_operation): Likewise.
7293 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7294 (vectorizable_bswap): Likewise.
7295 (vectorizable_call): Likewise.
7296 (vectorizable_conversion): Likewise.
7297 (vectorizable_assignment): Likewise.
7298 (vectorizable_shift): Likewise.
7299 (vectorizable_operation): Likewise.
7300 (vectorizable_store): Likewise.
7301 (vectorizable_load): Likewise.
7302 (vectorizable_condition): Likewise.
7303 (vectorizable_comparison): Likewise.
7304 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7306 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7307 Alan Hayward <alan.hayward@arm.com>
7308 David Sherwood <david.sherwood@arm.com>
7310 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7311 of vect_init_vector.
7313 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7314 Alan Hayward <alan.hayward@arm.com>
7315 David Sherwood <david.sherwood@arm.com>
7317 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7318 an inline wrapper that provides a location.
7319 (gimple_build_vector): Likewise.
7320 * gimple-fold.c (gimple_build_vector_from_val): New function.
7321 (gimple_build_vector): Likewise.
7322 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7323 functions to build the initial value. Always return a gimple value.
7324 (get_initial_defs_for_reduction): Likewise. Only compute
7326 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7327 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7328 (vectorizable_induction): Use gimple_build_vector rather than
7331 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7332 Alan Hayward <alan.hayward@arm.com>
7333 David Sherwood <david.sherwood@arm.com>
7335 * target.h (vec_perm_indices): New typedef.
7336 (auto_vec_perm_indices): Likewise.
7337 * optabs-query.h: Include target.h
7338 (can_vec_perm_p): Take a vec_perm_indices *.
7339 * optabs-query.c (can_vec_perm_p): Likewise.
7340 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7341 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7342 * tree-vect-generic.c (lower_vec_perm): Likewise.
7343 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7344 (vect_grouped_load_supported): Likewise.
7345 (vect_shift_permute_load_chain): Likewise.
7346 (vect_permute_store_chain): Use auto_vec_perm_indices.
7347 (vect_permute_load_chain): Likewise.
7348 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7349 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7350 Update uses of can_vec_perm_p.
7351 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7352 mode with a number of elements. Take a vec_perm_indices *.
7353 (vect_create_epilog_for_reduction): Update accordingly.
7354 Use auto_vec_perm_indices.
7355 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7356 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7357 (vect_transform_slp_perm_load): Likewise.
7358 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7359 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7360 (vect_gen_perm_mask_checked): Likewise.
7361 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7362 (vect_gen_perm_mask_checked): Likewise.
7363 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7364 (vectorizable_store): Likewise.
7365 (vectorizable_load): Likewise.
7366 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7367 (vectorizable_bswap): Likewise.
7369 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7370 Alan Hayward <alan.hayward@arm.com>
7371 David Sherwood <david.sherwood@arm.com>
7373 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7374 * tree.c (build_vector): Likewise.
7375 (build_vector_from_ctor): Update accordingly.
7376 (build_vector_from_val): Likewise.
7377 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7378 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7379 * tree-vect-generic.c (add_rshift): Likewise.
7380 (expand_vector_divmod): Likewise.
7381 (optimize_vector_constructor): Likewise.
7382 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7383 (vect_transform_slp_perm_load): Likewise.
7384 (vect_schedule_slp_instance): Likewise.
7385 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7386 (vectorizable_call): Likewise.
7387 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7388 * expmed.c (make_tree): Likewise.
7389 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7390 a vector passed to build_vector.
7391 (fold_convert_const): Likewise.
7392 (exact_inverse): Likewise.
7393 (fold_ternary_loc): Likewise.
7394 (fold_relational_const): Likewise.
7395 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7396 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7397 (const_unop): Likewise. Store the reduction accumulator in a
7398 variable rather than an array.
7399 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7400 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7401 the new vector, rather than constructing it after the input arrays.
7402 (native_interpret_vector): Use auto_vec<tree> when building
7403 a vector passed to build_vector. Add elements in order.
7404 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7405 auto_vec<tree> when building a vector passed to build_vector.
7406 (vect_create_epilog_for_reduction): Likewise.
7407 (vectorizable_induction): Likewise.
7408 (get_initial_def_for_reduction): Likewise. Fix indentation of
7410 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7412 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7413 passed to build_vector.
7415 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7416 Alan Hayward <alan.hayward@arm.com>
7417 David Sherwood <david.sherwood@arm.com>
7419 * tree-core.h (tree_base::u): Add an "nelts" field.
7420 (tree_vector): Use VECTOR_CST_NELTS as the length.
7421 * tree.c (tree_size): Likewise.
7422 (make_vector): Initialize VECTOR_CST_NELTS.
7423 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7424 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7425 TYPE_VECTOR_SUBPARTS.
7426 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7427 as the number of units, setting it from VECTOR_CST_NELTS.
7428 (const_vector_from_tree): Likewise.
7429 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7430 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7431 (fold_negate_expr_1): Likewise.
7432 (fold_convert_const): Likewise.
7433 (const_binop): Likewise. Differentiate the number of output and
7435 (const_unop): Likewise.
7436 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7437 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7438 in cases that did the opposite.
7440 2017-09-14 Richard Biener <rguenther@suse.de>
7442 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7445 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7447 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7449 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7452 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7453 if and where to split a bb, except for splitting before debug insn
7454 sequences followed by non-label real insn. Delete debug insns
7455 in between basic blocks.
7457 * combine.c (make_compound_operation_int): Formatting fixes.
7459 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7460 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7461 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7462 * config/sol2.h (LINK_EH_SPEC): Likewise.
7463 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7464 * config/s390/linux.h (LINK_SPEC): Likewise.
7465 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7466 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7467 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7468 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7469 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7470 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7471 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7472 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7474 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7476 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7478 (ENDFILE_LINUX_SPEC): Likewise.
7479 (LINK_EH_SPEC): Likewise.
7480 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7481 (LINK_OS_LINUX_SPEC32): Likewise.
7482 (LINK_OS_LINUX_SPEC64): Likewise.
7483 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7484 (LINK_OS_LINUX_SPEC): Likewise.
7486 2017-09-13 Martin Liska <mliska@suse.cz>
7489 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7490 CASE_HIGH is NULL_TREE.
7492 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7493 Alan Hayward <alan.hayward@arm.com>
7494 David Sherwood <david.sherwood@arm.com>
7496 * target.def (secondary_memory_needed): New hook.
7497 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7498 instead of SECONDARY_MEMORY_NEEDED.
7499 (secondary_memory_needed_mode): Likewise.
7500 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7501 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7502 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7503 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7504 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7505 * doc/tm.texi: Regenerate.
7506 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7507 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7508 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7509 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7510 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7511 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7512 mode argument first and change the reg_class arguments to reg_class_t.
7513 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7514 Make static. Update the call to inline_secondary_memory_needed.
7515 (ix86_register_move_cost): Update the call to
7516 inline_secondary_memory_needed.
7517 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7518 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7520 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7521 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7523 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7524 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7525 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7526 and match hook interface. Add comment from mips.h.
7527 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7528 * config/mmix/mmix.md (truncdfsf2): Refer to
7529 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7531 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7532 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7533 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7534 (pa_secondary_memory_needed): New function.
7535 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7536 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7537 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7538 (pdp11_secondary_memory_needed): Make static and match hook interface.
7539 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7540 * config/powerpcspe/powerpcspe-protos.h
7541 (rs6000_secondary_memory_needed_ptr): Delete.
7542 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7544 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7545 (rs6000_option_override_internal): Assign to
7546 targetm.secondary_memory_needed rather than
7547 rs6000_secondary_memory_needed_ptr.
7548 (rs6000_secondary_memory_needed): Match hook interface.
7549 (rs6000_debug_secondary_memory_needed): Likewise.
7550 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7551 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7552 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7553 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7554 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7555 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7557 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7558 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7559 (rs6000_option_override_internal): Assign to
7560 targetm.secondary_memory_needed rather than
7561 rs6000_secondary_memory_needed_ptr.
7562 (rs6000_secondary_memory_needed): Match hook interface.
7563 (rs6000_debug_secondary_memory_needed): Likewise.
7564 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7565 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7566 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7567 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7568 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7569 (sparc_secondary_memory_needed): New function.
7570 * lra-constraints.c (check_and_process_move): Refer to
7571 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7573 (curr_insn_transform): Likewise.
7574 (process_alt_operands): Use targetm.secondary_memory_needed
7575 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7576 (check_secondary_memory_needed_p): Likewise.
7577 (choose_split_class): Likewise.
7578 * reload.c: Unconditionally include code that was previously
7579 conditional on SECONDARY_MEMORY_NEEDED.
7580 (push_secondary_reload): Use targetm.secondary_memory_needed
7581 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7582 (push_reload): Likewise.
7583 * reload1.c: Unconditionally include code that was previously
7584 conditional on SECONDARY_MEMORY_NEEDED.
7585 (choose_reload_regs): Use targetm.secondary_memory_needed
7586 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7587 (gen_reload): Likewise.
7588 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7590 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7591 Alan Hayward <alan.hayward@arm.com>
7592 David Sherwood <david.sherwood@arm.com>
7594 * target.def (secondary_memory_needed_mode): New hook:
7595 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7596 * targhooks.h (default_secondary_memory_needed_mode): New function.
7597 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7598 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7599 * doc/tm.texi: Regenerate.
7600 * lra-constraints.c (check_and_process_move): Use
7601 targetm.secondary_memory_needed_mode instead of
7602 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7603 (curr_insn_transform): Likewise.
7604 * reload.c (get_secondary_mem): Likewise.
7605 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7606 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7608 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7609 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7610 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7611 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7612 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7614 * config/powerpcspe/powerpcspe-protos.h
7615 (rs6000_secondary_memory_needed_mode): Delete.
7616 * config/powerpcspe/powerpcspe.c
7617 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7618 (rs6000_secondary_memory_needed_mode): Make static.
7619 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7620 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7622 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7624 (rs6000_secondary_memory_needed_mode): Make static.
7625 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7626 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7627 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7628 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7629 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7631 (sparc_secondary_memory_needed_mode): New function.
7632 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7634 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7636 * config/aarch64/constraints.md (Umq): New constraint.
7637 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7639 (mov<mode>): Update condition.
7641 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7643 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7644 when bitposition is the same.
7646 2017-09-13 Richard Biener <rguenther@suse.de>
7648 * dwarf2out.c (output_die_symbol): Remove.
7649 (output_die): Do not output a DIEs symbol.
7651 2017-09-13 Richard Biener <rguenther@suse.de>
7654 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7655 default-def to avoid breaking iterator update with the weird
7656 interaction with cgraph_update_edges_for_call_stmt_node.
7658 2017-09-13 Richard Biener <rguenther@suse.de>
7660 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7661 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7662 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7663 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7665 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7667 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7668 Disable pc relative literal load irrespective of
7669 TARGET_FIX_ERR_A53_84341 for default.
7671 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7673 * config/sparc/sparc.c (output_return): Output the source location of
7674 the insn in the delay slot, if any.
7675 (output_sibcall): Likewise.
7677 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7680 * common.opt (-static-pie): New alias.
7681 (shared): Negate static-pie.
7682 (-no-pie): Update help text.
7684 (static-pie): New option.
7685 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7686 -static-pie support.
7687 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7688 (LINK_EH_SPEC): Likewise.
7689 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7690 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7691 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7692 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7693 (init_gcc_specs): Likewise.
7694 (init_spec): Likewise.
7695 (display_help): Update help message for -pie.
7696 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7699 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7701 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7702 (movdi_aarch64): Likewise.
7703 (movti_aarch64): Likewise.
7705 2017-09-12 Simon Wright <simon@pushface.org>
7708 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7709 calculation of the minor version, always output as 0.
7711 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7714 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7715 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7716 on it early, rather than manual conversion late. For
7717 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7718 instead of performing manual conversion.
7720 2017-09-12 Carl Love <cel@us.ibm.com>
7722 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7723 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7725 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7726 VMULOSW): Add definitions.
7727 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7728 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7729 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7731 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7733 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7735 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7737 (movti_aarch64): Likewise.
7738 (movdf_aarch64): Likewise.
7739 (movtf_aarch64): Likewise.
7740 (load_pairdi): Likewise.
7741 (store_pairdi): Likewise.
7742 (load_pairdf): Likewise.
7743 (store_pairdf): Likewise.
7744 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7745 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7746 (ldr_got_small_<mode>): Likewise.
7747 (ldr_got_small_28k_<mode>): Likewise.
7748 (ldr_got_tiny): Likewise.
7749 * config/aarch64/iterators.md (ldst_sz): New.
7750 (ldpstp_sz): Likewise.
7751 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7753 (thunderx_load): Split load_8 to load_16.
7754 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7756 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7757 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7758 (xgene1_store_pair): Split store_8 to store_16.
7759 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7760 (falkor_st_0_st_sd): Split store_8 to store_16.
7762 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7764 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7765 and store1/2/3/4 to store_4/8/12/16.
7766 * config/aarch64/aarch64.md: Update for rename.
7767 * config/arm/arm.md: Likewise.: Likewise.
7768 * config/arm/arm.c: Likewise.
7769 * config/arm/thumb1.md: Likewise.
7770 * config/arm/thumb2.md: Likewise.
7771 * config/arm/vfp.md: Likewise.
7772 * config/arm/arm-generic.md: Likewise.
7773 * config/arm/arm1020e.md: Likewise.
7774 * config/arm/arm1026ejs.md: Likewise.
7775 * config/arm/arm1136jfs.md: Likewise.
7776 * config/arm/arm926ejs.md: Likewise.
7777 * config/arm/cortex-a15.md: Likewise.
7778 * config/arm/cortex-a17.md: Likewise.
7779 * config/arm/cortex-a5.md: Likewise.
7780 * config/arm/cortex-a53.md: Likewise.
7781 * config/arm/cortex-a57.md: Likewise.
7782 * config/arm/cortex-a7.md: Likewise.
7783 * config/arm/cortex-a8.md: Likewise.
7784 * config/arm/cortex-a9.md: Likewise.
7785 * config/arm/cortex-m4.md: Likewise.
7786 * config/arm/cortex-m7.md: Likewise.
7787 * config/arm/cortex-r4.md: Likewise.
7788 * config/arm/exynos-m1.md: Likewise.
7789 * config/arm/fa526.md: Likewise.
7790 * config/arm/fa606te.md: Likewise.
7791 * config/arm/fa626te.md: Likewise.
7792 * config/arm/fa726te.md: Likewise.
7793 * config/arm/fmp626.md: Likewise.
7794 * config/arm/iwmmxt.md: Likewise.
7795 * config/arm/ldmstm.md: Likewise.
7796 * config/arm/marvell-pj4.md: Likewise.
7797 * config/arm/xgene1.md: Likewise.
7798 * config/aarch64/thunderx.md: Likewise.
7799 * config/aarch64/thunderx2t99.md: Likewise.
7800 * config/aarch64/falkor.md: Likewise.
7802 2017-09-12 Martin Liska <mliska@suse.cz>
7804 * attribs.c (private_lookup_attribute): New function.
7805 * attribs.h (private_lookup_attribute): Declared here.
7806 (lookup_attribute): Called from this place.
7808 2017-09-12 Richard Biener <rguenther@suse.de>
7810 PR tree-optimization/82157
7811 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7812 stmts with side-effects.
7814 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7815 Alan Hayward <alan.hayward@arm.com>
7816 David Sherwood <david.sherwood@arm.com>
7818 * target.def (hard_regno_nregs): New hook.
7819 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7820 * targhooks.h (default_hard_regno_nregs): Declare.
7821 * targhooks.c (default_hard_regno_nregs): New function.
7822 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7823 (TARGET_HARD_REGNO_NREGS): ...this hook.
7824 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7825 (CLASS_MAX_NREGS): Likewise.
7826 * doc/tm.texi: Regenerate.
7827 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7828 instead of HARD_REGNO_NREGS.
7829 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7830 HARD_REGNO_NREGS in the comment.
7831 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7832 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7833 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7834 Return an unsigned int.
7835 (TARGET_HARD_REGNO_NREGS): Redefine.
7836 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7837 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7838 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7839 (arc_hard_regno_nregs): New function.
7840 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7841 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7842 (arm_hard_regno_nregs): New function.
7843 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7844 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7845 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7846 (TARGET_HARD_REGNO_NREGS): Redefine.
7847 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7848 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7849 (HARD_REGNO_NREGS): Delete.
7850 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7851 (cr16_hard_regno_nregs): New function.
7852 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7853 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7854 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7855 (cris_hard_regno_nregs): New function.
7856 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7857 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7858 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7859 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7860 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7861 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7862 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7863 (frv_hard_regno_nregs): Make static. Take and return an
7865 (frv_class_max_nregs): Remove outdated copy of documentation.
7866 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7867 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7868 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7869 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7870 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7871 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7872 (TARGET_HARD_REGNO_NREGS): Redefine.
7873 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7874 (CLASS_MAX_NREGS): Update comment.
7875 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7876 (ia64_hard_regno_nregs): New function.
7877 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7878 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7879 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7880 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7881 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7883 (m32c_hard_regno_nregs): Likewise. Make static.
7884 (TARGET_HARD_REGNO_NREGS): Redefine.
7885 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7886 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7887 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7888 (m68k_hard_regno_nregs): New function.
7889 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7890 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7891 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7892 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7893 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7894 Take and return an unsigned int.
7895 (TARGET_HARD_REGNO_NREGS): Redefine.
7896 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7897 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7898 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7899 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7900 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7901 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7902 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7903 (msp430_hard_regno_nregs): Make static. Take and return an
7905 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7906 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7907 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7908 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7909 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7910 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7911 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7912 (TARGET_HARD_REGNO_NREGS): Redefine.
7913 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7914 (PA_HARD_REGNO_NREGS): ...this.
7915 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7916 (PA_HARD_REGNO_NREGS): ...this.
7917 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7918 (pa_hard_regno_nregs): New function.
7919 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7920 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7921 (pdp11_hard_regno_nregs): New function.
7922 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7923 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7924 (rs6000_hard_regno_nregs_hook): New function.
7925 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7926 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7927 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7928 Take and return an unsigned int. Move earlier in file.
7929 (TARGET_HARD_REGNO_NREGS): Redefine.
7930 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7931 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7932 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7933 (rl78_hard_regno_nregs): Make static. Take and return an
7935 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7936 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7937 (rs6000_hard_regno_nregs_hook): New function.
7938 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7939 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7940 (TARGET_HARD_REGNO_NREGS): Redefine.
7941 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7942 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7943 instead of HARD_REGNO_NREGS.
7944 (s390_hard_regno_nregs): New function.
7945 (s390_hard_regno_mode_ok): Add comment from s390.h.
7946 (TARGET_HARD_REGNO_NREGS): Redefine.
7947 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7948 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7949 (sh_hard_regno_nregs): New function.
7950 (sh_pass_in_reg_p): Use it.
7951 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7952 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7953 (sparc_hard_regno_nregs): New function.
7954 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7955 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7956 (spu_function_arg_advance): Use it, supplying a valid register number.
7957 (TARGET_HARD_REGNO_NREGS): Redefine.
7958 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7959 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7960 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7961 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7962 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7963 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7964 (CLASS_MAX_NREGS): Remove copy of old documentation.
7965 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7966 (visium_hard_regno_nregs): New function.
7967 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7968 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7969 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7970 xtensa_hard_regno_nregs): New function.
7971 * system.h (HARD_REGNO_NREGS): Poison.
7973 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7975 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7976 hard_regno_nregs instead of HARD_REGNO_NREGS.
7977 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7978 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7979 (c6x_expand_epilogue): Likewise.
7980 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7981 (frv_read_iacc_argument): Likewise.
7982 * config/sh/sh.c: Include regs.h.
7983 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7984 (regs_used): Likewise.
7985 (output_stack_adjust): Likewise.
7986 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7987 * expmed.c: Include regs.h.
7988 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7989 * ree.c: Include regs.h.
7990 (combine_reaching_defs): Use hard_regno_nregs instead of
7992 (add_removable_extension): Likewise.
7994 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7996 * regs.h (hard_regno_nregs): Turn into a function.
7997 (end_hard_regno): Update accordingly.
7998 * caller-save.c (setup_save_areas): Likewise.
7999 (save_call_clobbered_regs): Likewise.
8000 (replace_reg_with_saved_mem): Likewise.
8001 (insert_restore): Likewise.
8002 (insert_save): Likewise.
8003 * combine.c (can_change_dest_mode): Likewise.
8004 (move_deaths): Likewise.
8005 (distribute_notes): Likewise.
8006 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8007 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8008 (rs6000_split_multireg_move): Likewise.
8009 (rs6000_register_move_cost): Likewise.
8010 (rs6000_memory_move_cost): Likewise.
8011 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8012 (rs6000_split_multireg_move): Likewise.
8013 (rs6000_register_move_cost): Likewise.
8014 (rs6000_memory_move_cost): Likewise.
8015 * cselib.c (cselib_reset_table): Likewise.
8016 (cselib_lookup_1): Likewise.
8017 * emit-rtl.c (set_mode_and_regno): Likewise.
8018 * function.c (aggregate_value_p): Likewise.
8019 * ira-color.c (setup_profitable_hard_regs): Likewise.
8020 (check_hard_reg_p): Likewise.
8021 (calculate_saved_nregs): Likewise.
8022 (assign_hard_reg): Likewise.
8023 (improve_allocation): Likewise.
8024 (calculate_spill_cost): Likewise.
8025 * ira-emit.c (modify_move_list): Likewise.
8026 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8027 (ira_hard_reg_in_set_p): Likewise.
8028 * ira.c (setup_reg_mode_hard_regset): Likewise.
8029 (clarify_prohibited_class_mode_regs): Likewise.
8030 (check_allocation): Likewise.
8031 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8032 (lra_setup_reg_renumber): Likewise.
8033 (setup_try_hard_regno_pseudos): Likewise.
8034 (spill_for): Likewise.
8035 (assign_hard_regno): Likewise.
8036 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8037 * lra-constraints.c (in_class_p): Likewise.
8038 (lra_constraint_offset): Likewise.
8039 (simplify_operand_subreg): Likewise.
8040 (lra_constraints): Likewise.
8041 (split_reg): Likewise.
8042 (split_if_necessary): Likewise.
8043 (invariant_p): Likewise.
8044 (inherit_in_ebb): Likewise.
8045 * lra-lives.c (process_bb_lives): Likewise.
8046 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8047 (get_hard_regs): Likewise.
8048 (do_remat): Likewise.
8049 * lra-spills.c (assign_spill_hard_regs): Likewise.
8050 * mode-switching.c (create_pre_exit): Likewise.
8051 * postreload.c (reload_combine_recognize_pattern): Likewise.
8052 * recog.c (peep2_find_free_register): Likewise.
8053 * regcprop.c (kill_value_regno): Likewise.
8054 (set_value_regno): Likewise.
8055 (copy_value): Likewise.
8056 (maybe_mode_change): Likewise.
8057 (find_oldest_value_reg): Likewise.
8058 (copyprop_hardreg_forward_1): Likewise.
8059 * regrename.c (check_new_reg_p): Likewise.
8060 (regrename_do_replace): Likewise.
8061 * reload.c (push_reload): Likewise.
8062 (combine_reloads): Likewise.
8063 (find_dummy_reload): Likewise.
8064 (operands_match_p): Likewise.
8065 (find_reloads): Likewise.
8066 (find_equiv_reg): Likewise.
8067 (reload_adjust_reg_for_mode): Likewise.
8068 * reload1.c (count_pseudo): Likewise.
8069 (count_spilled_pseudo): Likewise.
8070 (find_reg): Likewise.
8071 (clear_reload_reg_in_use): Likewise.
8072 (free_for_value_p): Likewise.
8073 (allocate_reload_reg): Likewise.
8074 (choose_reload_regs): Likewise.
8075 (reload_adjust_reg_for_temp): Likewise.
8076 (emit_reload_insns): Likewise.
8077 (delete_output_reload): Likewise.
8078 * rtlanal.c (subreg_get_info): Likewise.
8079 * sched-deps.c (sched_analyze_reg): Likewise.
8080 * sel-sched.c (init_regs_for_mode): Likewise.
8081 (mark_unavailable_hard_regs): Likewise.
8082 (choose_best_reg_1): Likewise.
8083 (verify_target_availability): Likewise.
8084 * valtrack.c (dead_debug_insert_temp): Likewise.
8085 * var-tracking.c (track_loc_p): Likewise.
8086 (emit_note_insn_var_location): Likewise.
8087 * varasm.c (make_decl_rtl): Likewise.
8088 * reginfo.c (choose_hard_reg_mode): Likewise.
8089 (init_reg_modes_target): Refer directly to
8090 this_target_regs->x_hard_regno_nregs.
8092 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8094 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8095 instead of hard_regno_nregs.
8097 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8099 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8100 end_hard_regno instead of hard_regno_nregs.
8101 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8102 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8103 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8104 * ira-color.c (improve_allocation): Likewise.
8105 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8106 * lra-lives.c (mark_regno_live): Likewise.
8107 (mark_regno_dead): Likewise.
8108 * lra-remat.c (operand_to_remat): Likewise.
8109 * lra.c (collect_non_operand_hard_regs): Likewise.
8110 * postreload.c (reload_combine_note_store): Likewise.
8111 (move2add_valid_value_p): Likewise.
8112 * reload.c (regno_clobbered_p): Likewise.
8114 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8116 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8118 * config/v850/v850.c (v850_reorg): Likewise.
8119 * reload.c (refers_to_regno_for_reload_p): Likewise.
8120 (find_equiv_reg): Likewise.
8121 * reload1.c (reload_reg_reaches_end_p): Likewise.
8123 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8125 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8127 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8128 * config/arm/arm.c (output_move_neon): Likewise.
8129 (arm_attr_length_move_neon): Likewise.
8130 (neon_split_vcombine): Likewise.
8131 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8132 (c6x_mark_reg_written): Likewise.
8133 (c6x_dwarf_register_span): Likewise.
8134 * config/i386/i386.c (ix86_save_reg): Likewise.
8135 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8136 (rws_access_reg): Likewise.
8137 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8138 * mode-switching.c (create_pre_exit): Likewise.
8139 * ree.c (combine_reaching_defs): Likewise.
8140 (add_removable_extension): Likewise.
8141 * regcprop.c (find_oldest_value_reg): Likewise.
8142 (copyprop_hardreg_forward_1): Likewise.
8143 * reload.c (reload_inner_reg_of_subreg): Likewise.
8144 (push_reload): Likewise.
8145 (combine_reloads): Likewise.
8146 (find_dummy_reload): Likewise.
8147 (reload_adjust_reg_for_mode): Likewise.
8148 * reload1.c (find_reload_regs): Likewise.
8149 (forget_old_reloads_1): Likewise.
8150 (reload_reg_free_for_value_p): Likewise.
8151 (reload_adjust_reg_for_temp): Likewise.
8152 (emit_reload_insns): Likewise.
8153 (delete_output_reload): Likewise.
8154 * sel-sched.c (choose_best_reg_1): Likewise.
8155 (choose_best_pseudo_reg): Likewise.
8157 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8158 Alan Hayward <alan.hayward@arm.com>
8159 David Sherwood <david.sherwood@arm.com>
8161 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8162 * target.def (slow_unaligned_access): New hook.
8163 * targhooks.h (default_slow_unaligned_access): Declare.
8164 * targhooks.c (default_slow_unaligned_access): New function.
8165 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8166 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8167 * doc/tm.texi: Regenerate.
8168 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8169 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8170 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8172 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8173 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8175 (rs6000_slow_unaligned_access): New function.
8176 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8177 (expand_block_compare): Likewise.
8178 (expand_strn_compare): Likewise.
8179 (rs6000_rtx_costs): Likewise.
8180 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8181 (riscv_slow_unaligned_access): Likewise.
8182 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8183 (riscv_slow_unaligned_access_p): ...this and make static.
8184 (riscv_option_override): Update accordingly.
8185 (riscv_slow_unaligned_access): New function.
8186 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8187 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8188 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8189 (rs6000_slow_unaligned_access): New function.
8190 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8191 (rs6000_rtx_costs): Likewise.
8192 * config/rs6000/rs6000-string.c (expand_block_compare)
8193 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8194 of SLOW_UNALIGNED_ACCESS.
8195 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8196 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8197 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8198 of SLOW_UNALIGNED_ACCESS.
8199 * expmed.c (simple_mem_bitfield_p): Likewise.
8200 * expr.c (alignment_for_piecewise_move): Likewise.
8201 (emit_group_load_1): Likewise.
8202 (emit_group_store): Likewise.
8203 (copy_blkmode_from_reg): Likewise.
8204 (emit_push_insn): Likewise.
8205 (expand_assignment): Likewise.
8206 (store_field): Likewise.
8207 (expand_expr_real_1): Likewise.
8208 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8209 * lra-constraints.c (simplify_operand_subreg): Likewise.
8210 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8211 * gimple-ssa-store-merging.c: Likewise in block comment at start
8213 * tree-ssa-strlen.c: Include target.h.
8214 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8215 of SLOW_UNALIGNED_ACCESS.
8216 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8218 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8220 PR rtl-optimization/82185
8221 * expmed.c (emit_store_flag_int): Only test tem if it has been
8224 2017-09-12 Richard Biener <rguenther@suse.de>
8227 * match.pd ((FTYPE) N CMP CST): Fix typo.
8229 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8231 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8233 (mips_near_type_p): Add 'short_call' attribute as a synonym
8235 * doc/extend.texi (short_call): Document new function attribute.
8237 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8240 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8241 assertion check that in the condition.
8242 (get_atomic_generic_size): Likewise. Before testing if parameter
8243 has pointer type, if it has array type, call for C++
8244 default_conversion to perform array-to-pointer conversion.
8246 2017-09-12 Richard Biener <rguenther@suse.de>
8248 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8249 for operations we cannot scalarize.
8251 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8253 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8254 vectors heap vectors. Clean up comments.
8255 Make visited_bbs a reference.
8256 (profitable_jump_thread_path): Make GC
8257 vectors heap vectors. Clean up comments.
8259 (convert_and_register_jump_thread_path): Make GC vectors heap
8261 (check_subpath_and_update_thread_path): Same. Clean up comments.
8262 Make visited_bbs a reference.
8263 (handle_phi): Abstract common code to to
8264 register_jump_thread_path_if_profitable.
8265 Rename VAR_BB to DEF_BB.
8267 Make GC vectors heap vectors.
8268 Make visited_bbs a reference.
8269 (handle_assignment): Same.
8270 (register_jump_thread_path_if_profitable): New.
8271 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8273 Make GC vectors heap vectors. Clean up comments.
8274 Make visited_bbs a reference.
8275 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8276 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8279 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8282 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8283 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8285 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8287 Revert r251800 and r251799.
8289 2017-09-11 Martin Jambor <mjambor@suse.cz>
8292 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8293 arguments in advance.
8294 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8295 use it to find predecessor edges.
8296 (naive_outof_ssa): Collect vector of predecessors.
8298 2017-09-08 Jason Merrill <jason@redhat.com>
8300 PR c++/70029 - ICE with ref-qualifier and -flto
8301 * langhooks.h (struct lang_hooks_for_types): Add
8302 copy_lang_qualifiers.
8303 * attribs.c (build_type_attribute_qual_variant): Use it.
8304 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8306 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8307 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8309 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8312 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8313 (*mulsi3_sp64): New instruction.
8314 (mulsi3): New expander.
8316 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8318 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8320 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8322 * sancov.c: Include memmodel.h.
8324 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8327 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8330 2017-09-07 Carl Love <cel@us.ibm.com>
8332 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8333 the sldi instruction.
8335 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8337 * sancov.c: Include tm_p.h.
8339 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8342 * output.h (switch_to_other_text_partition): New declaration.
8343 * varasm.c (switch_to_other_text_partition): New function.
8344 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8345 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8346 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8347 to the other text partition before emitting LCL label and switch back
8348 after emitting the word after it.
8350 2017-09-07 Richard Biener <rguenther@suse.de>
8352 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8353 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8354 critical edges here, after loop init.
8355 (pass_data_pre): Remove PROP_no_crit_edges flags.
8356 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8357 for valueization of call args to avoid leaking VN_TOP.
8358 (visit_use): Assert we do not visit default defs.
8359 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8360 Use error_mark_node to more easily detect leaking VN_TOP.
8361 All default-defs are varying, not VN_TOP. Mark them visited.
8362 (run_scc_vn): Make code match comment.
8364 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8366 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8367 OPTION_MASK_FLOAT128_KEYWORD.
8368 (POWERPC_MASKS): Likewise.
8369 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8370 support for the -mfloat128-type option, and make -mfloat128
8371 default on PowerPC Linux systems. Define or undefine
8372 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8373 Define __float128 to be __ieee128 if IEEE 128-bit support is
8374 enabled, or undefine it.
8375 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8376 Delete defining __FLOAT128_TYPE__.
8377 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8378 -mfloat128-type option and make -mfloat128 default on PowerPC
8380 (TARGET_FLOAT128_TYPE): Likewise.
8381 (-mfloat128-type): Likewise.
8382 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8383 Delete the -mfloat128-type option and make -mfloat128 default on
8384 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8385 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8386 128-bit floating point is enabled. Change tests from using
8387 -mfloat128-type to -mfloat128.
8388 (rs6000_mangle_type): Use the correct mangling for the __float128
8389 type even if normal long double is restricted to 64-bits.
8390 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8392 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8393 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8394 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8396 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8397 documentation for -mfloat128.
8399 2017-09-06 Olivier Hainque <hainque@adacore.com>
8401 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8403 2017-09-06 Wish Wu <wishwu007@gmail.com>
8404 Jakub Jelinek <jakub@redhat.com>
8406 * asan.c (initialize_sanitizer_builtins): Add
8407 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8408 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8409 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8410 BT_FN_VOID_UINT64_PTR variables.
8411 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8412 (BT_FN_VOID_UINT16_UINT16): Likewise.
8413 (BT_FN_VOID_UINT32_UINT32): Likewise.
8414 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8415 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8416 (BT_FN_VOID_UINT64_PTR): Likewise.
8417 * common.opt (flag_sanitize_coverage): New variable.
8418 (fsanitize-coverage=trace-pc): Remove.
8419 (fsanitize-coverage=): Add.
8420 * flag-types.h (enum sanitize_coverage_code): New enum.
8421 * fold-const.c (fold_range_test): Disable non-short-circuit
8422 optimization if flag_sanitize_coverage.
8423 (fold_truth_andor): Likewise.
8424 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8425 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8426 (coverage_sanitizer_opts): New array.
8427 (get_closest_sanitizer_option): Add OPTS argument, handle also
8428 OPT_fsanitize_coverage_.
8429 (parse_sanitizer_options): Adjusted to also handle
8430 OPT_fsanitize_coverage_.
8431 (common_handle_option): Add OPT_fsanitize_coverage_.
8432 * sancov.c (instrument_comparison, instrument_switch): New function.
8433 (sancov_pass): Add trace-cmp support.
8434 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8435 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8436 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8437 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8438 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8439 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8440 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8441 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8442 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8443 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8445 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8447 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8448 error. Only quit immediately if parsing is complete.
8449 (BEGIN): Initialize fatal_err and parse_done.
8450 (begin fpu, end fpu): Check number of arguments.
8451 (begin arch, end arch): Likewise.
8452 (begin cpu, end cpu): Likewise.
8453 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8454 (optalias): Likewise.
8456 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8458 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8459 * config/arm/arm-isa.h: Delete. Move definitions to ...
8460 * arm-cpus.in: ... here. Use new feature and fgroup values.
8461 * config/arm/arm.c (arm_option_override): Use lower case for feature
8463 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8464 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8465 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8467 (print_isa_bits_for): New function.
8468 (gen_isa): New function.
8469 (gen_comm_data): Use print_isa_bits_for.
8470 (define feature): New keyword.
8471 (define fgroup): New keyword.
8472 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8473 (arm-isa.h): Add rule to generate file.
8474 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8475 case for feature bit names.
8477 2017-09-06 Richard Biener <rguenther@suse.de>
8479 * tree-ssa-pre.c (NECESSARY): Remove.
8480 (create_expression_by_pieces): Do not touch pass-local flags.
8481 (insert_into_preds_of_block): Likewise.
8482 (do_pre_regular_insertion): Likewise.
8483 (eliminate_insert): Likewise.
8484 (eliminate_dom_walker::before_dom_children): Likewise.
8485 (fini_eliminate): Do not look at inserted_exprs.
8486 (mark_operand_necessary): Remove.
8487 (remove_dead_inserted_code): Replace with simple work-list
8488 algorithm based on inserted_exprs and SSA uses.
8489 (pass_pre::execute): Re-order fini_eliminate and
8490 remove_dead_inserted_code.
8492 2017-09-06 Olivier Hainque <hainque@adacore.com>
8494 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8495 for VxWorks 7. Adjust surrounding comments.
8497 2017-09-06 Richard Biener <rguenther@suse.de>
8499 * gimple-ssa-strength-reduction.c
8500 (find_candidates_dom_walker::before_dom_children): Also allow
8503 2017-09-06 Richard Biener <rguenther@suse.de>
8505 PR tree-optimization/82108
8506 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8507 for gap in the non-permutation SLP case.
8509 2017-09-06 Martin Jambor <mjambor@suse.cz>
8511 PR tree-optimization/82078
8512 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8513 add_access_to_work_queue...
8514 (build_accesses_from_assign): ...here.
8515 (propagate_all_subaccesses): Make sure racc is the group
8516 representative, if there is one.
8518 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8521 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8524 2017-09-06 Richard Biener <rguenther@suse.de>
8526 * gimple-ssa-strength-reduction.c
8527 (find_candidates_dom_walker::before_doom_children): Use a
8528 type and not a mode check.
8530 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8533 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8535 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8537 2017-09-05 Jeff Law <law@redhat.com>
8539 PR tree-optimization/64910
8540 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8541 cases where we have 3 or more operands.
8543 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8546 * omp-low.c (lower_omp_for): Recompute tree invariant if
8547 gimple_omp_for_initial/final is ADDR_EXPR.
8550 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8551 into gimple val before gimplification fo the COND_EXPR.
8553 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8555 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8556 REGION_COPY argument.
8557 (thread_through_all_blocks): Remove unused argument to
8558 duplicate_thread_path.
8560 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8561 Alan Hayward <alan.hayward@arm.com>
8562 David Sherwood <david.sherwood@arm.com>
8564 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8565 Take a scalar_mode rather than a machine_mode.
8566 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8567 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8568 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8569 (aarch64_gen_adjusted_ldpstp): Likewise.
8570 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8572 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8573 Alan Hayward <alan.hayward@arm.com>
8574 David Sherwood <david.sherwood@arm.com>
8576 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8577 Take a scalar_int_mode instead of a machine_mode.
8578 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8579 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8580 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8581 (aarch64_simd_attr_length_rglist): Delete.
8582 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8583 a scalar_int_mode instead of a machine_mode.
8584 (aarch64_add_offset): Likewise.
8585 (aarch64_internal_mov_immediate): Likewise
8586 (aarch64_add_constant_internal): Likewise.
8587 (aarch64_add_constant): Likewise.
8588 (aarch64_movw_imm): Likewise.
8589 (aarch64_rtx_arith_op_extract_p): Likewise.
8590 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8591 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8592 Remove assert that the mode isn't a vector.
8593 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8594 (aarch64_expand_mov_immediate): Update calls after above changes.
8595 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8596 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8597 (aarch64_move_imm): Likewise.
8598 (aarch64_can_const_movi_rtx_p): Likewise.
8599 (aarch64_strip_extend): Likewise.
8600 (aarch64_extr_rtx_p): Likewise.
8601 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8602 a CONST_INT when the mode parameter is VOIDmode.
8603 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8605 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8607 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8608 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8609 (bitwise_type_for_mode): Update accordingly.
8611 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8613 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8614 * stor-layout.c (mode_for_size_tree): Likewise.
8615 (mode_for_array): Update accordingly.
8616 (layout_decl): Likewise.
8617 (compute_record_mode): Likewise. Only set the mode once.
8619 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8621 * target.def (get_mask_mode): Change return type to opt_mode.
8623 * doc/tm.texi: Regenerate.
8624 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8625 * targhooks.c (default_get_mask_mode): Likewise.
8626 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8627 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8628 targetm.get_mask_mode.
8629 * tree.c (build_truth_vector_type): Likewise.
8631 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8633 * machmode.h (mode_for_vector): Return an opt_mode.
8634 * stor-layout.c (mode_for_vector): Likewise.
8635 (mode_for_int_vector): Update accordingly.
8636 (layout_type): Likewise.
8637 * config/i386/i386.c (emit_memmov): Likewise.
8638 (ix86_expand_set_or_movmem): Likewise.
8639 (ix86_expand_vector_init): Likewise.
8640 (ix86_get_mask_mode): Likewise.
8641 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8643 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8644 * expmed.c (extract_bit_field_1): Likewise.
8645 * expr.c (expand_expr_real_2): Likewise.
8646 * optabs-query.c (can_vec_perm_p): Likewise.
8647 (can_vec_mask_load_store_p): Likewise.
8648 * optabs.c (expand_vec_perm): Likewise.
8649 * targhooks.c (default_get_mask_mode): Likewise.
8650 * tree-vect-stmts.c (vectorizable_store): Likewise.
8651 (vectorizable_load): Likewise.
8652 (get_vectype_for_scalar_type_and_size): Likewise.
8654 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8656 * machmode.h (mode_for_int_vector): New function.
8657 * stor-layout.c (mode_for_int_vector): Likewise.
8658 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8659 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8660 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8661 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8662 (s390_expand_vcond): Likewise.
8664 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8666 * machmode.h (opt_machine_mode): New type.
8667 (opt_mode<T>): Allow construction from anything that can be
8669 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8670 (mode_for_size): Return an opt_machine_mode.
8671 * stor-layout.c (mode_for_size): Likewise.
8672 (mode_for_size_tree): Update call accordingly.
8673 (bitwise_mode_for_mode): Likewise.
8674 (make_fract_type): Likewise.
8675 (make_accum_type): Likewise.
8676 * caller-save.c (replace_reg_with_saved_mem): Update call
8678 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8679 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8680 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8681 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8682 * expmed.c (extract_bit_field_1): Likewise.
8683 * reload.c (get_secondary_mem): Likewise.
8684 * varasm.c (assemble_integer): Likewise.
8685 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8688 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8690 * machmode.h (decimal_float_mode_for_size): New function.
8691 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8692 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8693 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8694 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8695 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8696 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8698 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8700 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8701 (get_builtin_sync_mode): Likewise.
8702 (expand_ifn_atomic_compare_exchange): Likewise.
8703 (expand_builtin_atomic_clear): Likewise.
8704 (expand_builtin_atomic_test_and_set): Likewise.
8705 (fold_builtin_atomic_always_lock_free): Likewise.
8706 * calls.c (compute_argument_addresses): Likewise.
8707 (emit_library_call_value_1): Likewise.
8708 (store_one_arg): Likewise.
8709 * combine.c (combine_instructions): Likewise.
8710 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8711 * config/arm/arm.c (arm_function_value): Likewise.
8712 (aapcs_allocate_return_reg): Likewise.
8713 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8714 * config/i386/i386.c (construct_container): Likewise.
8715 (ix86_gimplify_va_arg): Likewise.
8716 (ix86_expand_sse_cmp): Likewise.
8717 (emit_memmov): Likewise.
8718 (emit_memset): Likewise.
8719 (expand_small_movmem_or_setmem): Likewise.
8720 (ix86_expand_pextr): Likewise.
8721 (ix86_expand_pinsr): Likewise.
8722 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8723 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8725 * config/mips/mips.c (mips_function_value_1) Likewise.
8726 (mips_block_move_straight): Likewise.
8727 (mips_expand_ins_as_unaligned_store): Likewise.
8728 * config/powerpcspe/powerpcspe.c
8729 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8730 (rs6000_darwin64_record_arg_flush): Likewise.
8731 * config/rs6000/rs6000.c
8732 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8733 (rs6000_darwin64_record_arg_flush): Likewise.
8734 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8735 (sparc_function_value_1): Likewise.
8736 * config/spu/spu.c (adjust_operand): Likewise.
8737 (spu_emit_branch_or_set): Likewise.
8738 (arith_immediate_p): Likewise.
8739 * emit-rtl.c (gen_lowpart_common): Likewise.
8740 * expr.c (expand_expr_real_1): Likewise.
8741 * function.c (assign_parm_setup_block): Likewise.
8742 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8743 * reload1.c (alter_reg): Likewise.
8744 * stor-layout.c (mode_for_vector): Likewise.
8745 (layout_type): Likewise.
8747 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8749 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8750 (spu_convert_move): Likewise.
8751 * lower-subreg.c (resolve_simple_move): Likewise.
8753 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8756 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8757 define_insn to a define_expand.
8758 (altivec_vsum2sws_direct): New define_insn.
8759 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8761 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8763 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8766 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8771 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8772 realign stack if stack alignment needed is less than incoming
8775 2017-09-05 Marek Polacek <polacek@redhat.com>
8778 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8781 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8783 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8785 2017-09-05 Richard Biener <rguenther@suse.de>
8787 PR tree-optimization/82084
8788 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8790 2017-09-05 Richard Biener <rguenther@suse.de>
8792 PR tree-optimization/82102
8793 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8795 2017-09-05 Martin Liska <mliska@suse.cz>
8797 PR tree-optimization/82032
8798 * tree-cfg.c (generate_range_test): New function.
8799 * tree-cfg.h (generate_range_test): Declared here.
8800 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8801 (cleanup_control_expr_graph): Use it.
8802 * tree-switch-conversion.c (try_switch_expansion): Remove
8804 (emit_case_nodes): Use generate_range_test.
8806 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8809 * config/i386/i386.md (*<btsc><mode>_mask): Add
8810 TARGET_USE_BT to insn constraint.
8811 (*btr<mode>_mask): Ditto.
8813 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8815 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8816 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8818 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8821 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8822 TARGET_NEON and TARGET_IWMMXT.
8823 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8824 TARGET_NEON and TARGET_IWMMXT.
8825 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8827 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8829 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8830 (ix86_rewrite_tls_address): Ditto.
8831 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8832 (ix86_rewrite_tls_address_1): Ditto.
8833 (ix86_rewrite_tls_address): Ditto.
8834 * config/i386/predicates.md (tls_address_pattern): New predicate.
8835 * config/i386/i386.md (TLS address splitter): New splitter.
8837 2017-09-04 Richard Biener <rguenther@suse.de>
8839 PR tree-optimization/82084
8840 * fold-const.h (can_native_encode_string_p): Declare.
8841 * fold-const.c (can_native_encode_string_p): Factor out from ...
8842 (native_encode_string): ... here.
8843 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8844 vectorizing stores from constants we later cannot handle.
8846 2017-09-04 Marek Polacek <polacek@redhat.com>
8849 * doc/invoke.texi: Update -Wtautological-compare documentation.
8851 2017-09-04 Jeff Law <law@redhat.com>
8853 PR tree-optimization/64910
8854 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8855 swap the first and last operand if the last is a constant.
8857 2017-09-04 Marek Polacek <polacek@redhat.com>
8860 * convert.c (do_narrow): When sanitizing signed integer overflows,
8861 bail out for signed types.
8862 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8864 2017-09-04 Richard Biener <rguenther@suse.de>
8866 PR tree-optimization/82060
8867 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8868 Move devirtualization after stmt folding and before EH/AB/noreturn
8869 cleanup to get the stmt refs canonicalized. Use a bool instead
8870 of gimple_modified_p since that doesn't work for NOPs. Schedule
8871 NOPs generated by folding for removal.
8873 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8874 Alan Hayward <alan.hayward@arm.com>
8875 David Sherwood <david.sherwood@arm.com>
8877 * coretypes.h (pad_direction): New enum.
8878 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8879 (FUNCTION_ARG_PADDING): Likewise.
8880 * target.def (function_arg_padding): New hook.
8881 * targhooks.h (default_function_arg_padding): Declare.
8882 * targhooks.c (default_function_arg_padding): New function.
8883 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8884 (TARGET_FUNCTION_ARG_PADDING): ...this.
8885 * doc/tm.texi: Regenerate.
8886 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8887 instead of direction.
8888 (compute_argument_addresses): Likewise.
8889 (load_register_parameters): Likewise.
8890 (emit_library_call_value_1): Likewise.
8891 (store_one_arg): Use targetm.calls.function_arg_padding instead
8892 of FUNCTION_ARG_PADDING.
8893 (must_pass_in_stack_var_size_or_pad): Likewise.
8894 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8895 (emit_group_store): Likewise.
8896 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8897 instead of FUNCTION_ARG_PADDING.
8898 (emit_push_insn): Likewise, and propagate enum change throughout
8900 * function.h (direction): Delete.
8901 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8903 * function.c (assign_parm_find_stack_rtl): Likewise.
8904 (assign_parm_setup_block_p): Likewise.
8905 (assign_parm_setup_block): Likewise.
8906 (gimplify_parameters): Likewise.
8907 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8908 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8910 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8911 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8912 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8913 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8914 (aarch64_function_arg_padding): ...this new function.
8915 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8916 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8917 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8918 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8919 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8920 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8921 (arm_pad_arg_upward): Replace with...
8922 (arm_function_arg_padding): ...this new function.
8923 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8925 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8926 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8927 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8928 (ia64_hpux_function_arg_padding): Replace with...
8929 (ia64_function_arg_padding): ...this new function. Use pad_direction
8930 instead of direction. Check for TARGET_HPUX.
8931 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8932 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8933 (iq2000_function_arg_padding): New function.
8934 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8935 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8936 (mips_function_arg_padding): ...this new function.
8937 (mips_pad_reg_upward): Update accordingly.
8938 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8939 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8940 targetm.calls.function_arg_padding.
8941 (FUNCTION_ARG_PADDING): Delete.
8942 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8943 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8944 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8945 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8946 (nios2_block_reg_padding): Return pad_direction instead of direction.
8947 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8948 instead of direction.
8949 (nios2_function_arg_padding): Likewise. Make static.
8950 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8951 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8952 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8953 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8954 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8955 (pa_function_arg_padding): Make static. Return pad_direction instead
8957 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8958 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8959 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8960 instead of direction. Use targetm.calls.function_arg_padding.
8961 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8962 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8963 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8964 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8965 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8967 (function_arg_padding): Rename to...
8968 (rs6000_function_arg_padding): ...this. Make static. Return
8969 pad_direction instead of direction.
8970 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8971 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8972 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8973 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8974 instead of direction. Use targetm.calls.function_arg_padding.
8975 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8976 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8977 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8978 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8979 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8980 (function_arg_padding): Rename to...
8981 (rs6000_function_arg_padding): ...this. Make static. Return
8982 pad_direction instead of direction.
8983 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8984 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8985 * config/s390/s390.c (s390_function_arg_padding): New function.
8986 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8987 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8988 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8989 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8990 (function_arg_padding): Rename to...
8991 (sparc_function_arg_padding): ...this. Make static. Return
8992 pad_direction instead of direction.
8993 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8994 * config/spu/spu.c (spu_function_arg_padding): New function.
8995 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8996 * system.h (FUNCTION_ARG_PADDING): Poison.
8998 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8999 Alan Hayward <alan.hayward@arm.com>
9000 David Sherwood <david.sherwood@arm.com>
9002 * target.def (modes_tieable_p): New hook.
9003 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9004 (TARGET_MODES_TIEABLE_P): ...this.
9005 * doc/tm.texi.in: Regenerate.
9006 * hooks.h (hook_bool_mode_mode_true): Declare.
9007 * hooks.c (hook_bool_mode_mode_true): New function.
9008 * combine.c (subst): Use targetm.modes_tieable_p instead of
9010 * dse.c (find_shift_sequence): Likewise.
9011 * expmed.c (extract_low_bits): Likewise.
9012 * lower-subreg.c: Include target.h.
9013 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9015 * rtlanal.c (rtx_cost): Likewise.
9016 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9017 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9018 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9019 (TARGET_MODES_TIEABLE_P): Redefine.
9020 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9021 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9022 (TARGET_MODES_TIEABLE_P): Redefine.
9023 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9024 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9025 (arc_modes_tieable_p): New function.
9026 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9027 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9028 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9029 (arm_modes_tieable_p): Make static.
9030 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9031 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9032 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9033 (TARGET_MODES_TIEABLE_P): Redefine.
9034 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9035 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9036 (TARGET_MODES_TIEABLE_P): Redefine.
9037 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9038 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9039 (cr16_modes_tieable_p): New function.
9040 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9041 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9042 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9043 (TRULY_NOOP_TRUNCATION): Update comment.
9044 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9045 (TRULY_NOOP_TRUNCATION): Update comment.
9046 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9047 (frv_modes_tieable_p): New function.
9048 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9049 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9050 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9051 (TARGET_MODES_TIEABLE_P): Redefine.
9052 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9053 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9054 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9055 (TARGET_MODES_TIEABLE_P): Redefine.
9056 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9057 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9058 (ia64_modes_tieable_p): New function.
9059 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9060 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9061 (iq2000_modes_tieable_p): New function.
9062 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9063 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9064 (lm32_modes_tieable_p): New function.
9065 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9066 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9067 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9068 (TARGET_MODES_TIEABLE_P): Redefine.
9069 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9070 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9071 (m32r_modes_tieable_p): New function.
9072 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9073 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9074 (m68k_modes_tieable_p): New function.
9075 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9076 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9077 (mcore_modes_tieable_p): New function.
9078 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9079 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9081 (TARGET_MODES_TIEABLE_P): Redefine.
9082 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9083 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9084 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9085 (TARGET_MODES_TIEABLE_P): Redefine.
9086 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9087 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9088 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9089 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9090 (mn10300_modes_tieable_p): ...this and make static.
9091 (TARGET_MODES_TIEABLE_P): Redefine.
9092 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9093 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9094 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9095 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9096 (msp430_modes_tieable_p): Make static.
9097 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9098 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9099 (TARGET_MODES_TIEABLE_P): Redefine.
9100 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9101 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9102 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9103 (TARGET_MODES_TIEABLE_P): Redefine.
9104 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9105 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9106 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9107 (TARGET_MODES_TIEABLE_P): Redefine.
9108 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9109 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9110 (pdp11_modes_tieable_p): New function.
9111 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9112 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9113 (rs6000_modes_tieable_p): New function.
9114 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9115 * config/powerpcspe/powerpcspe.md: Update comment.
9116 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9117 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9118 (TARGET_MODES_TIEABLE_P): Redefine.
9119 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9120 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9121 (rl78_modes_tieable_p): New function.
9122 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9123 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9124 (rs6000_modes_tieable_p): New function.
9125 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9126 * config/rs6000/rs6000.md: Update comment.
9127 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9128 * config/rx/rx.c (rx_modes_tieable_p): New function.
9129 (TARGET_MODES_TIEABLE_P): Redefine.
9130 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9131 * config/s390/s390.c (s390_modes_tieable_p): New function.
9132 (TARGET_MODES_TIEABLE_P): Redefine.
9133 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9134 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9135 (sh_modes_tieable_p): New function.
9136 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9137 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9138 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9139 (sparc_modes_tieable_p): Make static.
9140 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9141 * config/spu/spu.c (spu_modes_tieable_p): New function.
9142 (TARGET_MODES_TIEABLE_P): Redefine.
9143 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9144 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9145 (TARGET_MODES_TIEABLE_P): Redefine.
9146 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9147 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9148 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9149 * config/v850/v850.c (v850_modes_tieable_p): New function.
9150 (TARGET_MODES_TIEABLE_P): Redefine.
9151 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9152 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9153 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9154 (visium_modes_tieable_p): New function.
9155 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9156 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9157 (xtensa_modes_tieable_p): New function.
9158 * system.h (MODES_TIEABLE_P): Poison.
9160 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9161 Alan Hayward <alan.hayward@arm.com>
9162 David Sherwood <david.sherwood@arm.com>
9164 * target.def (hard_regno_mode_ok): New hook.
9165 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9166 (TARGET_HARD_REGNO_MODE_OK): ...this.
9167 * doc/tm.texi.in: Regenerate.
9168 * hooks.h (hook_bool_uint_mode_true): Declare.
9169 * hooks.c (hook_bool_uint_mode_true): New function.
9170 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9172 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9173 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9174 instead of HARD_REGNO_MODE_OK.
9175 * caller-save.c: Include target.h.
9176 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9178 * combine.c (can_combine_p): Likewise.
9179 (combinable_i3pat): Likewise.
9180 (can_change_dest_mode): Likewise.
9181 * expr.c (init_expr_target): Likewise.
9182 (convert_move): Likewise.
9183 (convert_modes): Likewise.
9184 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9185 (setup_prohibited_mode_move_regs): Likewise.
9186 * ira.h (target_ira): Likewise.
9187 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9188 * lra-constraints.c (process_alt_operands): Likewise.
9189 (split_reg): Likewise.
9190 * recog.c (peep2_find_free_register): Likewise.
9191 * ree.c (combine_reaching_defs): Likewise.
9192 * regcprop.c (maybe_mode_change): Likewise.
9193 * reginfo.c (init_reg_sets_1): Likewise.
9194 (choose_hard_reg_mode): Likewise.
9195 (simplifiable_subregs): Likewise.
9196 * regrename.c (check_new_reg_p): Likewise.
9197 * reload.c (find_valid_class): Likewise.
9198 (find_valid_class_1): Likewise.
9199 (reload_inner_reg_of_subreg): Likewise.
9200 (push_reload): Likewise.
9201 (combine_reloads): Likewise.
9202 (find_dummy_reload): Likewise.
9203 (find_reloads): Likewise.
9204 * reload1.c (find_reg): Likewise.
9205 (set_reload_reg): Likewise.
9206 (allocate_reload_reg): Likewise.
9207 (choose_reload_regs): Likewise.
9208 (reload_adjust_reg_for_temp): Likewise.
9209 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9210 (simplify_subreg_regno): Likewise.
9211 * sel-sched.c (init_regs_for_mode): Likewise.
9212 * varasm.c (make_decl_rtl): Likewise.
9213 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9214 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9216 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9217 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9218 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9219 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9220 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9221 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9222 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9223 (arc_mode_class): Delete.
9224 (HARD_REGNO_MODE_OK): Delete.
9225 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9226 (arc_hard_regno_mode_ok): Rename old array to...
9227 (arc_hard_regno_mode_ok_modes): ...this.
9228 (arc_conditional_register_usage): Update accordingly.
9229 (arc_mode_class): Make static.
9230 (arc_hard_regno_mode_ok): New function.
9231 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9232 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9233 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9234 (arm_hard_regno_mode_ok): Make static.
9235 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9237 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9238 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9239 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9242 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9243 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9244 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9245 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9247 * config/bfin/predicates.md (valid_reg_operand): Use
9248 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9249 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9250 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9251 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9252 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9253 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9254 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9255 (cr16_hard_regno_mode_ok): Make static and return a bool.
9256 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9257 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9258 (cris_hard_regno_mode_ok): New function.
9259 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9260 (epiphany_mode_class): Delete.
9261 (HARD_REGNO_MODE_OK): Delete.
9262 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9263 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9264 (hard_regno_mode_ok): Rename to...
9265 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9266 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9267 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9269 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9270 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9271 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9272 (frv_hard_regno_mode_ok): Make static and return a bool.
9273 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9275 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9276 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9277 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9278 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9281 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9282 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9283 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9285 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9286 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9287 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9288 (ia64_hard_regno_mode_ok): New function.
9289 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9290 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9291 (iq2000_hard_regno_mode_ok): New function.
9292 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9293 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9294 (lm32_hard_regno_mode_ok): New function.
9295 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9296 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9297 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9298 instead of HARD_REGNO_MODE_OK.
9299 (m32c_hard_regno_ok): Rename to...
9300 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9301 (m32c_cannot_change_mode_class): Update accordingly.
9302 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9303 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9304 (m32r_mode_class): Delete.
9305 (HARD_REGNO_MODE_OK): Delete.
9306 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9307 (m32r_hard_regno_mode_ok): Rename to...
9308 (m32r_hard_regno_modes): ...this.
9309 (m32r_mode_class): Make static.
9310 (m32r_hard_regno_mode_ok): New function.
9311 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9312 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9313 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9314 (m68k_hard_regno_mode_ok): Make static.
9315 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9316 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9317 (mcore_hard_regno_mode_ok): New function.
9318 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9319 (HARD_REGNO_MODE_OK): Delete.
9320 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9322 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9323 (microblaze_hard_regno_mode_ok): New function.
9324 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9325 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9326 (mips_hard_regno_mode_ok): Delete.
9327 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9328 (mips_hard_regno_mode_ok_p): ...this and make static.
9329 (mips_hard_regno_mode_ok_p): Rename to...
9330 (mips_hard_regno_mode_ok_uncached): ...this.
9331 (mips_hard_regno_mode_ok): New function.
9332 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9333 of HARD_REGNO_MODE_OK.
9334 (mips_option_override): Update after above name changes.
9335 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9336 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9337 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9338 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9339 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9340 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9341 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9342 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9343 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9344 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9345 (msp430_hard_regno_mode_ok): Make static and return a bool.
9346 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9347 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9348 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9350 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9351 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9352 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9353 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9354 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9355 (PA_HARD_REGNO_MODE_OK): ...this
9356 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9357 (PA_HARD_REGNO_MODE_OK): ...this.
9358 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9359 (pa_hard_regno_mode_ok): New function.
9360 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9361 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9362 (pdp11_hard_regno_mode_ok): New function.
9363 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9364 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9366 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9368 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9369 (rs6000_hard_regno_mode_ok): Rename to...
9370 (rs6000_hard_regno_mode_ok_uncached): ...this.
9371 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9372 (rs6000_hard_regno_mode_ok): New function.
9373 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9374 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9375 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9376 (riscv_hard_regno_mode_ok): ...this and make static.
9377 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9378 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9379 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9380 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9381 (rl78_hard_regno_mode_ok): Make static and return bool.
9382 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9383 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9385 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9386 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9387 (rs6000_hard_regno_mode_ok): Rename to...
9388 (rs6000_hard_regno_mode_ok_uncached): ...this.
9389 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9390 (rs6000_hard_regno_mode_ok): New function.
9391 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9392 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9393 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9394 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9395 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9396 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9398 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9399 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9400 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9401 (sh_hard_regno_mode_ok): Make static.
9402 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9403 instead of HARD_REGNO_MODE_OK.
9404 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9405 (sparc_mode_class): Delete.
9406 (HARD_REGNO_MODE_OK): Delete.
9407 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9408 (hard_regno_mode_classes): Make static.
9409 (sparc_mode_class): Likewise.
9410 (sparc_hard_regno_mode_ok): New function.
9411 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9412 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9413 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9415 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9416 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9417 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9418 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9419 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9421 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9422 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9423 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9424 (visium_hard_regno_mode_ok): New function.
9425 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9426 instead of HARD_REGNO_MODE_OK.
9427 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9428 (HARD_REGNO_MODE_OK): Delete.
9429 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9430 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9431 (xtensa_option_override): Update accordingly.
9432 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9433 (xtensa_hard_regno_mode_ok): New function.
9434 * system.h (HARD_REGNO_MODE_OK): Poison.
9436 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9437 Alan Hayward <alan.hayward@arm.com>
9438 David Sherwood <david.sherwood@arm.com>
9440 * target.def (hard_regno_call_part_clobbered): New hook.
9441 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9442 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9443 * doc/tm.texi: Regenerate.
9444 * hooks.h (hook_bool_uint_mode_false): Declare.
9445 * hooks.c (hook_bool_uint_mode_false): New function.
9446 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9447 * cselib.c (cselib_process_insn): Use
9448 targetm.hard_regno_call_part_clobbered instead of
9449 HARD_REGNO_CALL_PART_CLOBBERED.
9450 * ira-conflicts.c (ira_build_conflicts): Likewise.
9451 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9452 * lra-constraints.c (need_for_call_save_p): Likewise.
9453 * lra-lives.c: Include target.h.
9454 (check_pseudos_live_through_calls): Use
9455 targetm.hard_regno_call_part_clobbered instead of
9456 HARD_REGNO_CALL_PART_CLOBBERED.
9457 * regcprop.c: Include target.h.
9458 (copyprop_hardreg_forward_1): Use
9459 targetm.hard_regno_call_part_clobbered instead of
9460 HARD_REGNO_CALL_PART_CLOBBERED.
9461 * reginfo.c (choose_hard_reg_mode): Likewise.
9462 * regrename.c (check_new_reg_p): Likewise.
9463 * reload.c (find_equiv_reg): Likewise.
9464 * reload1.c (emit_reload_insns): Likewise.
9465 * sched-deps.c (deps_analyze_insn): Likewise.
9466 * sel-sched.c (init_regs_for_mode): Likewise.
9467 (mark_unavailable_hard_regs): Likewise.
9468 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9469 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9470 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9472 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9473 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9474 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9476 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9478 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9479 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9480 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9482 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9483 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9484 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9486 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9487 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9489 * config/powerpcspe/powerpcspe.c
9490 (rs6000_hard_regno_call_part_clobbered): New function.
9491 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9492 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9493 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9495 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9496 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9497 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9499 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9500 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9501 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9503 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9504 Alan Hayward <alan.hayward@arm.com>
9505 David Sherwood <david.sherwood@arm.com>
9507 * rtl.h (subreg_memory_offset): Declare.
9508 * emit-rtl.c (subreg_memory_offset): New function.
9509 * expmed.c (store_bit_field_1): Use it.
9510 * expr.c (undefined_operand_subword_p): Likewise.
9511 * simplify-rtx.c (simplify_subreg): Likewise.
9513 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9515 PR rtl-optimization/57448
9518 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9519 using atomic_load pattern.
9520 (expand_atomic_store): Likewise.
9522 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9525 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9526 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9527 handling. Use replace_call_with_value with NULL instead of
9528 gsi_replace, unlink_stmt_vdef and release_defs.
9530 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9533 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9535 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9538 * rtl.h (emit_library_call_value_1): Declare.
9539 (emit_library_call): Replace declaration with a series of overloads.
9540 Remove the parameter count argument.
9541 (emit_library_call_value): Likewise.
9542 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9543 with an "rtx_mode_t *".
9544 (emit_library_call_value): Delete.
9545 (emit_library_call): Likewise.
9546 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9547 (asan_emit_allocas_unpoison): Likewise.
9548 * builtins.c (expand_builtin_powi): Likewise.
9549 (expand_asan_emit_allocas_unpoison): Likewise.
9550 * cfgexpand.c (expand_main_function): Likewise.
9551 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9552 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9553 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9554 * config/arm/arm.c (arm_trampoline_init): Likewise.
9555 (arm_call_tls_get_addr): Likewise.
9556 (arm_expand_divmod_libfunc): Likewise.
9557 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9558 (smulsi3_highpart): Likewise.
9559 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9560 (c6x_expand_compare): Likewise.
9561 (c6x_expand_movmem): Likewise.
9562 * config/frv/frv.c (frv_trampoline_init): Likewise.
9563 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9564 (ix86_expand_divmod_libfunc): Likewise.
9565 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9566 (ia64_expand_compare): Likewise.
9567 (ia64_profile_hook): Likewise.
9568 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9569 (nonlocal_goto): Likewise.
9570 (restore_stack_nonlocal): Likewise.
9571 * config/m32r/m32r.c (block_move_call): Likewise.
9572 (m32r_trampoline_init): Likewise.
9573 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9574 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9575 (m68k_call_m68k_read_tp): Likewise.
9576 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9577 (microblaze_expand_divide): Likewise.
9578 * config/mips/mips.h (mips_args): Likewise.
9579 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9580 (MIPS_ICACHE_SYNC): Likewise.
9581 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9582 (nios2_trampoline_init): Likewise.
9583 * config/pa/pa.c (hppa_tls_call): Likewise.
9584 (pa_trampoline_init): Likewise.
9585 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9586 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9587 (expand_strn_compare): Likewise.
9588 (rs6000_generate_compare): Likewise.
9589 (rs6000_expand_float128_convert): Likewise.
9590 (output_profile_hook): Likewise.
9591 (rs6000_trampoline_init): Likewise.
9592 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9593 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9594 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9595 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9596 (rs6000_generate_compare): Likewise.
9597 (rs6000_expand_float128_convert): Likewise.
9598 (output_profile_hook): Likewise.
9599 (rs6000_trampoline_init): Likewise.
9600 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9601 * config/sh/sh.c (sh_trampoline_init): Likewise.
9602 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9603 (sparc_emit_float_lib_cmp): Likewise.
9604 (sparc32_initialize_trampoline): Likewise.
9605 (sparc64_initialize_trampoline): Likewise.
9606 (sparc_profile_hook): Likewise.
9607 * config/spu/spu.c (ea_load_store): Likewise.
9608 * config/spu/spu.md (floatunssidf2): Likewise.
9609 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9610 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9611 * config/visium/visium.c (expand_block_move_4): Likewise.
9612 (expand_block_move_2): Likewise.
9613 (expand_block_move_1): Likewise.
9614 (expand_block_set_4): Likewise.
9615 (expand_block_set_2): Likewise.
9616 (expand_block_set_1): Likewise.
9617 (visium_trampoline_init): Likewise.
9618 (visium_profile_hook): Likewise.
9619 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9620 (xtensa_setup_frame_addresses): Likewise.
9621 (xtensa_trampoline_init): Likewise.
9622 * except.c (sjlj_emit_function_enter): Likewise.
9623 (sjlj_emit_function_exit): Likewise.
9624 * explow.c (allocate_dynamic_stack_space): Likewise.
9625 (probe_stack_range): Likewise.
9626 * expr.c (convert_mode_scalar): Likewise.
9627 * optabs.c (expand_binop): Likewise.
9628 (expand_twoval_binop_libfunc): Likewise.
9629 (expand_unop): Likewise.
9630 (prepare_cmp_insn): Likewise.
9631 (prepare_float_lib_cmp): Likewise.
9632 (expand_float): Likewise.
9633 (expand_fix): Likewise.
9634 (expand_fixed_convert): Likewise.
9635 (maybe_emit_sync_lock_test_and_set): Likewise.
9636 (expand_atomic_compare_and_swap): Likewise.
9637 (expand_mem_thread_fence): Likewise.
9638 (expand_atomic_fetch_op): Likewise.
9640 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9642 * doc/generic.texi (OpenACC): Adjust URL.
9643 * doc/invoke.texi (C Dialect Options): Ditto.
9645 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9647 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9648 predicate for operand 1. Add (m,<S>) constraint.
9649 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9650 Prevent memory operand 1 with register operand 2.
9652 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9654 PR rtl-optimization/82024
9655 * combine.c (try_combine): If the combination result is a PARALLEL,
9656 and we only need to retain the SET in there that would be placed
9657 at I2, check that we can place that at I3 instead, before doing so.
9659 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9662 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9664 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9665 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9668 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9669 Jeff Law <law@redhat.com>
9671 * varasm.c (bss_initializer_p): Do not put constants into .bss
9672 (categorize_decl_for_section): Handle bss_initializer_p returning
9673 false when DECL_INITIAL is NULL.
9675 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9678 * config/s390/s390.c (s390_can_inline_p): New function.
9680 2017-09-01 Jeff Law <law@redhat.com>
9682 PR tree-optimization/82052
9683 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9684 Always initialize the returned slot after a hash table miss
9685 when INSERT is true.
9687 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9689 * config/s390/s390.md (mem_signal_fence): Remove.
9690 * doc/md.texi (mem_signal_fence): Remove.
9691 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9693 * target-insns.def (mem_signal_fence): Remove.
9695 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9698 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9701 * asan.c (create_odr_indicator): Strip name encoding from assembler
9702 name before appending it after __odr_asan_.
9704 2017-09-01 Martin Liska <mliska@suse.cz>
9706 PR tree-optimization/82059
9707 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9708 frequency only when an edge is redirected.
9710 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9712 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9713 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9714 (arc_conditional_register_usage): Remove ARC600 lp_count
9716 (arc_file_start): Emit Tag_ARC_CPU_variation.
9717 (arc_can_use_doloop_p): New conditions to use ZOLs.
9718 (hwloop_fail): New function.
9719 (hwloop_optimize): Likewise.
9720 (hwloop_pattern_reg): Likewise.
9721 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9722 (arc_reorg_loops): New function, calls reorg_loops.
9723 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9724 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9726 (arc_loop_hazard): Remove function, functionality moved into
9728 (arc_hazard): Remove arc_loop_hazard call.
9729 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9730 into hwloop_optimize.
9731 (arc_label_align): Remove ZOL handling.
9732 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9733 * config/arc/arc.md (doloop_begin): Remove pattern.
9734 (doloop_begin_i): Likewise.
9735 (doloop_end_i): Likewise.
9736 (doloop_fallback): Likewise.
9737 (doloop_fallback_m): Likewise.
9738 (doloop_end): Reimplement expand.
9739 (arc_lp): New pattern for LP instruction.
9740 (loop_end): New pattern.
9741 (loop_fail): Likewise.
9742 (decrement_and_branch_until_zero): Likewise.
9743 * config/arc/arc.opt (mlpc-width): New option.
9744 * doc/invoke.texi (mlpc-width): Document option.
9746 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9748 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9749 (arc_ccfsm_advance): Fix checking for delay slots.
9750 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9752 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9754 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9756 (movhi_insn): Update store instruction constraint which are saving
9757 6-bit short immediates.
9758 (movsi_insn): Consider also short scaled load operations.
9759 (zero_extendhisi2_i): Use Usd constraint instead of T.
9760 (extendhisi2_i): Add q constraint.
9761 (arc_clzsi2): Add type and length attributes.
9762 (arc_ctzsi2): Likewise.
9763 * config/arc/constraints.md (Usc): Update constraint, the
9764 assembler can parse two relocations for a single instruction.
9766 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9768 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9769 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9771 2017-08-31 Olivier Hainque <hainque@adacore.com>
9773 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9774 match as powerpc-wrs-vxworks*.
9776 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9778 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9779 register constraint for by-element operand.
9780 (aarch64_mls_elt_merge<mode>): Likewise.
9782 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9784 * config/arc/arc.c (arc_can_follow_jump): Check for short
9787 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9789 * config.gcc: Use g.opt for arc.
9790 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9791 functionality moved to ...
9792 (legitimate_scaled_address_p): New function, ...here.
9793 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9794 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9795 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9797 (arc_override_options): Handle G option.
9798 (arc_output_pic_addr_const): Correct function definition.
9799 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9800 (arc_decl_anon_ns_mem_p): Delete.
9801 (arc_in_small_data_p): Overhaul this function to take into
9802 consideration the value given via G option.
9803 (arc_rewrite_small_data_1): Renamed and corrected old
9804 arc_rewrite_small_data function.
9805 (arc_rewrite_small_data): New function.
9806 (small_data_pattern): Don't use pic_offset_table_rtx.
9807 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9808 * config/arc/simdext.md (movmisalignv2hi): Use
9809 prepare_move_operands function.
9811 (movmisalign*): Likewise.
9812 * doc/invoke.texi (ARC options): Document -G option.
9814 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9816 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9818 * config/arc/arc.c (arc_print_operand): Output scalled address for
9819 sdata whenever is possible.
9820 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9821 load/stores are available.
9822 (compact_sda_memory_operand): Check for the alignment required by
9823 code density instructions.
9824 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9826 * config/arc/constraints.md (Usd): Update constraint.
9827 (Us0): New constraint.
9828 (Usc): Update constraint.
9830 2017-08-31 Richard Biener <rguenther@suse.de>
9833 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9836 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9838 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9839 Resize type_signature.
9841 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9842 Alan Hayward <alan.hayward@arm.com>
9843 David Sherwood <david.sherwood@arm.com>
9845 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9846 subregs whose inner modes can be stored in GPRs.
9847 (aarch64_classify_index): Likewise.
9849 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9850 Alan Hayward <alan.hayward@arm.com>
9851 David Sherwood <david.sherwood@arm.com>
9853 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9854 (V_INT_EQUIV): ...this.
9855 (v_cmp_result): Rename to...
9856 (v_int_equiv): ...this.
9857 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9858 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9859 (copysign<mode>3): Likewise.
9860 (aarch64_simd_bsl<mode>_internal): Likewise.
9861 (aarch64_simd_bsl<mode>): Likewise.
9862 (vec_cmp<mode><mode>): Likewise.
9863 (vcond<mode><mode>): Likewise.
9864 (vcond<v_cmp_mixed><mode>): Likewise.
9865 (vcondu<mode><v_cmp_mixed>): Likewise.
9866 (aarch64_cm<optab><mode>): Likewise.
9867 (aarch64_cmtst<mode>): Likewise.
9868 (aarch64_fac<optab><mode>): Likewise.
9869 (vec_perm_const<mode>): Likewise.
9870 (vcond_mask_<mode><v_cmp_result>): Rename to...
9871 (vcond_mask_<mode><v_int_equiv>): ...this.
9872 (vec_cmp<mode><v_cmp_result>): Rename to...
9873 (vec_cmp<mode><v_int_equiv>): ...this.
9875 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9876 Alan Hayward <alan.hayward@arm.com>
9877 David Sherwood <david.sherwood@arm.com>
9879 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9881 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9882 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9883 (UNSPEC_LD4_DREG): New unspecs.
9884 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9885 (aarch64_ld2<mode>_dreg_be): Replace with...
9886 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9888 (aarch64_ld3<mode>_dreg_le)
9889 (aarch64_ld3<mode>_dreg_be): Replace with...
9890 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9892 (aarch64_ld4<mode>_dreg_le)
9893 (aarch64_ld4<mode>_dreg_be): Replace with...
9894 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9897 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9899 PR tree-optimization/81987
9900 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9901 insert an initializer in a location not dominated by the stride
9904 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9906 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9907 on the entire header of the finally block in the fallthru case.
9909 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9911 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9913 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9915 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9916 rs6000_emit_move_from_cr and call renamed function.
9917 (rs6000_emit_prologue): Call renamed functions.
9918 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9919 movesi_from_cr, remove volatile CRs.
9921 2017-08-30 Jon Beniston <jon@beniston.com>
9922 Richard Biener <rguenther@suse.de>
9924 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9925 of VECTOR_MODE_P check.
9926 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9927 element vector types.
9929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9931 * df.h (df_read_modify_subreg_p): Remove in favor of...
9932 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9933 const_rtx instead of an rtx.
9934 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9935 * df-problems.c (df_word_lr_mark_ref): Likewise.
9936 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9937 (mark_pseudo_reg_dead): Likewise.
9938 (mark_ref_dead): Likewise.
9939 * reginfo.c (init_subregs_of_mode): Likewise.
9940 * sched-deps.c (sched_analyze_1): Likewise.
9941 * df-scan.c (df_def_record_1): Likewise.
9942 (df_uses_record): Likewise.
9943 (df_read_modify_subreg_p): Remove in favor of...
9944 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9945 const_rtx instead of an rtx.
9947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9948 Alan Hayward <alan.hayward@arm.com>
9949 David Sherwood <david.sherwood@arm.com>
9951 * rtl.h (partial_subreg_p): New function.
9952 * caller-save.c (save_call_clobbered_regs): Use it.
9953 * calls.c (expand_call): Likewise.
9954 * combine.c (combinable_i3pat): Likewise.
9955 (simplify_set): Likewise.
9956 (make_extraction): Likewise.
9957 (make_compound_operation_int): Likewise.
9958 (gen_lowpart_or_truncate): Likewise.
9959 (force_to_mode): Likewise.
9960 (make_field_assignment): Likewise.
9961 (reg_truncated_to_mode): Likewise.
9962 (record_truncated_value): Likewise.
9963 (move_deaths): Likewise.
9964 * cse.c (record_jump_cond): Likewise.
9965 (cse_insn): Likewise.
9966 * cselib.c (cselib_lookup_1): Likewise.
9967 * expmed.c (extract_bit_field_using_extv): Likewise.
9968 * function.c (assign_parm_setup_reg): Likewise.
9969 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9970 * ira-build.c (create_insn_allocnos): Likewise.
9971 * lra-coalesce.c (merge_pseudos): Likewise.
9972 * lra-constraints.c (match_reload): Likewise.
9973 (simplify_operand_subreg): Likewise.
9974 (curr_insn_transform): Likewise.
9975 * lra-lives.c (process_bb_lives): Likewise.
9976 * lra.c (new_insn_reg): Likewise.
9977 (lra_substitute_pseudo): Likewise.
9978 * regcprop.c (mode_change_ok): Likewise.
9979 (maybe_mode_change): Likewise.
9980 (copyprop_hardreg_forward_1): Likewise.
9981 * reload.c (push_reload): Likewise.
9982 (find_reloads): Likewise.
9983 (find_reloads_subreg_address): Likewise.
9984 * reload1.c (alter_reg): Likewise.
9985 (eliminate_regs_1): Likewise.
9986 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9988 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9990 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9991 back to if statements, including unpack.
9993 2017-08-30 Martin Liska <mliska@suse.cz>
9996 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9998 (func_checker::compare_asm_inputs_outputs): ... this function.
9999 (func_checker::compare_gimple_asm): Use the function to compare
10000 also ASM constrains.
10001 * ipa-icf-gimple.h: Rename the function.
10003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10004 Alan Hayward <alan.hayward@arm.com>
10005 David Sherwood <david.sherwood@arm.com>
10007 * coretypes.h (complex_mode): New type.
10008 * gdbhooks.py (build_pretty_printer): Handle it.
10009 * machmode.h (complex_mode): New class.
10010 (complex_mode::includes_p): New function.
10011 (is_complex_int_mode): Likewise.
10012 (is_complex_float_mode): Likewise.
10013 * genmodes.c (get_mode_class): Handle complex mode classes.
10014 * function.c (expand_function_end): Use is_complex_int_mode.
10016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10017 Alan Hayward <alan.hayward@arm.com>
10018 David Sherwood <david.sherwood@arm.com>
10020 * coretypes.h (scalar_mode_pod): New typedef.
10021 * gdbhooks.py (build_pretty_printer): Handle it.
10022 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10023 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10024 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10025 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10026 as_a <scalar_mode>.
10028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10029 Alan Hayward <alan.hayward@arm.com>
10030 David Sherwood <david.sherwood@arm.com>
10032 * machmode.h (mode_for_vector): Take a scalar_mode instead
10034 * stor-layout.c (mode_for_vector): Likewise.
10035 * explow.c (promote_mode): Use as_a <scalar_mode>.
10036 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10039 Alan Hayward <alan.hayward@arm.com>
10040 David Sherwood <david.sherwood@arm.com>
10042 * target.def (preferred_simd_mode): Take a scalar_mode
10043 instead of a machine_mode.
10044 * targhooks.h (default_preferred_simd_mode): Likewise.
10045 * targhooks.c (default_preferred_simd_mode): Likewise.
10046 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10047 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10048 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10049 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10050 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10051 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10052 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10053 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10055 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10056 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10057 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10058 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10059 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10060 * doc/tm.texi: Regenerate.
10061 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10065 Alan Hayward <alan.hayward@arm.com>
10066 David Sherwood <david.sherwood@arm.com>
10068 * target.def (scalar_mode_supported_p): Take a scalar_mode
10069 instead of a machine_mode.
10070 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10071 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10072 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10073 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10074 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10075 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10076 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10077 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10078 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10079 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10080 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10081 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10082 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10083 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10085 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10086 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10087 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10088 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10089 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10091 * doc/tm.texi: Regenerate.
10093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10094 Alan Hayward <alan.hayward@arm.com>
10095 David Sherwood <david.sherwood@arm.com>
10097 * coretypes.h (opt_scalar_mode): New typedef.
10098 * gdbhooks.py (build_pretty_printers): Handle it.
10099 * machmode.h (mode_iterator::get_2xwider): Add overload for
10101 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10103 * expr.c (convert_mode_scalar): Likewise.
10104 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10105 * optabs.c (expand_float): Likewise.
10106 (expand_fix): Likewise.
10107 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10110 Alan Hayward <alan.hayward@arm.com>
10111 David Sherwood <david.sherwood@arm.com>
10113 * optabs.c (expand_float): Explicitly check for scalars before
10114 using a branching expansion.
10115 (expand_fix): Likewise.
10117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10118 Alan Hayward <alan.hayward@arm.com>
10119 David Sherwood <david.sherwood@arm.com>
10121 * expr.c (convert_mode): Split scalar handling out into...
10122 (convert_mode_scalar): ...this new function. Treat the modes
10125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10126 Alan Hayward <alan.hayward@arm.com>
10127 David Sherwood <david.sherwood@arm.com>
10129 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10131 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10134 Alan Hayward <alan.hayward@arm.com>
10135 David Sherwood <david.sherwood@arm.com>
10137 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10138 rather than a machine_mode.
10139 (fixed_from_string): Likewise.
10140 (fixed_convert): Likewise.
10141 (fixed_convert_from_int): Likewise.
10142 (fixed_convert_from_real): Likewise.
10143 (real_convert_from_fixed): Likewise.
10144 * fixed-value.c (fixed_from_double_int): Likewise.
10145 (fixed_from_string): Likewise.
10146 (fixed_convert): Likewise.
10147 (fixed_convert_from_int): Likewise.
10148 (fixed_convert_from_real): Likewise.
10149 (real_convert_from_fixed): Likewise.
10150 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10153 Alan Hayward <alan.hayward@arm.com>
10154 David Sherwood <david.sherwood@arm.com>
10156 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10157 of separate mode class checks. Do not allow vector modes here.
10158 (immed_wide_int_const): Use as_a <scalar_mode>.
10159 * explow.c (trunc_int_for_mode): Likewise.
10160 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10161 (wi::shwi): Likewise.
10162 (wi::min_value): Likewise.
10163 (wi::max_value): Likewise.
10164 * dwarf2out.c (loc_descriptor): Likewise.
10165 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10166 for CONST_WIDE_INT.
10168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10169 Alan Hayward <alan.hayward@arm.com>
10170 David Sherwood <david.sherwood@arm.com>
10172 * tree.h (SCALAR_TYPE_MODE): New macro.
10173 * expr.c (expand_expr_addr_expr_1): Use it.
10174 (expand_expr_real_2): Likewise.
10175 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10176 (fold_convert_const_fixed_from_int): Likewise.
10177 (fold_convert_const_fixed_from_real): Likewise.
10178 (native_encode_fixed): Likewise
10179 (native_encode_complex): Likewise
10180 (native_encode_vector): Likewise.
10181 (native_interpret_fixed): Likewise.
10182 (native_interpret_real): Likewise.
10183 (native_interpret_complex): Likewise.
10184 (native_interpret_vector): Likewise.
10185 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10186 (simd_clone_adjust_argument_types): Likewise.
10187 (simd_clone_init_simd_arrays): Likewise.
10188 (simd_clone_adjust): Likewise.
10189 * stor-layout.c (layout_type): Likewise.
10190 * tree.c (build_minus_one_cst): Likewise.
10191 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10192 * tree-inline.c (estimate_move_cost): Likewise.
10193 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10194 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10195 (vectorizable_reduction): Likewise.
10196 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10197 (vect_recog_mixed_size_cond_pattern): Likewise.
10198 (check_bool_pattern): Likewise.
10199 (adjust_bool_pattern): Likewise.
10200 (search_type_for_mask_1): Likewise.
10201 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10202 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10203 (vectorizable_load): Likewise.
10204 (vectorizable_store): Likewise.
10205 * ubsan.c (ubsan_encode_value): Likewise.
10206 * varasm.c (output_constant): Likewise.
10208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10209 Alan Hayward <alan.hayward@arm.com>
10210 David Sherwood <david.sherwood@arm.com>
10212 * coretypes.h (scalar_mode): New class.
10213 * machmode.h (scalar_mode): Likewise.
10214 (scalar_mode::includes_p): New function.
10215 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10216 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10217 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10218 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10219 * expmed.c (store_bit_field_1): Likewise.
10220 (extract_bit_field_1): Likewise.
10221 * expr.c (write_complex_part): Likewise.
10222 (read_complex_part): Likewise.
10223 (emit_move_complex_push): Likewise.
10224 (expand_expr_real_2): Likewise.
10225 * function.c (assign_parm_setup_reg): Likewise.
10226 (assign_parms_unsplit_complex): Likewise.
10227 * optabs.c (expand_binop): Likewise.
10228 * rtlanal.c (subreg_get_info): Likewise.
10229 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10230 * varasm.c (output_constant_pool_2): Likewise.
10232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10233 Alan Hayward <alan.hayward@arm.com>
10234 David Sherwood <david.sherwood@arm.com>
10236 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10238 (expmed_mult_highpart_optab): Likewise.
10239 (expmed_mult_highpart): Likewise.
10241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10242 Alan Hayward <alan.hayward@arm.com>
10243 David Sherwood <david.sherwood@arm.com>
10245 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10246 instead of a machine_mode.
10247 (builtin_memset_read_str): Likewise.
10248 * builtins.c (c_readstr): Likewise.
10249 (builtin_memcpy_read_str): Likewise.
10250 (builtin_strncpy_read_str): Likewise.
10251 (builtin_memset_read_str): Likewise.
10252 (builtin_memset_gen_str): Likewise.
10253 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10254 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10255 instead of a machine_mode.
10256 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10258 (make_extraction): Likewise.
10259 (try_widen_shift_mode): Take and return scalar_int_modes instead
10261 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10262 a scalar_int_mode instead of a machine_mode.
10263 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10264 (avr_addr_space_pointer_mode): Likewise.
10265 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10266 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10267 (msp430_unwind_word_mode): Likewise.
10268 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10269 (spu_addr_space_pointer_mode): Likewise.
10270 (spu_addr_space_address_mode): Likewise.
10271 (spu_libgcc_cmp_return_mode): Likewise.
10272 (spu_libgcc_shift_count_mode): Likewise.
10273 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10274 (rl78_addr_space_pointer_mode): Likewise.
10275 (fl78_unwind_word_mode): Likewise.
10276 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10278 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10279 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10280 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10281 (mips_valid_pointer_mode): Likewise.
10282 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10283 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10284 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10286 (ft32_addr_space_address_mode): Likewise.
10287 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10288 scalar_int_mode instead of a machine_mode.
10289 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10291 (m32c_addr_space_address_mode): Likewise.
10292 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10293 (rs6000_eh_return_filter_mode): Likewise.
10294 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10295 (rs6000_eh_return_filter_mode): Likewise.
10296 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10297 (s390_libgcc_shift_count_mode): Likewise.
10298 (s390_unwind_word_mode): Likewise.
10299 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10301 * target.def (mode_rep_extended): Likewise.
10302 (valid_pointer_mode): Likewise.
10303 (addr_space.valid_pointer_mode): Likewise.
10304 (eh_return_filter_mode): Return a scalar_int_mode rather than
10306 (libgcc_cmp_return_mode): Likewise.
10307 (libgcc_shift_count_mode): Likewise.
10308 (unwind_word_mode): Likewise.
10309 (addr_space.pointer_mode): Likewise.
10310 (addr_space.address_mode): Likewise.
10311 * doc/tm.texi: Regenerate.
10312 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10314 (do_jump): Use scalar_int_mode for local variables.
10315 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10316 rather than a machine_mode.
10317 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10318 (scompare_loc_descriptor_wide): Likewise.
10319 (scompare_loc_descriptor_narrow): Likewise.
10320 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10322 * except.c (sjlj_emit_dispatch_table): Likewise.
10323 (expand_builtin_eh_copy_values): Likewise.
10324 * explow.c (convert_memory_address_addr_space_1): Likewise.
10325 Take a scalar_int_mode rather than a machine_mode.
10326 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10327 than a machine_mode.
10328 (memory_address_addr_space): Use scalar_int_mode for local variables.
10329 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10330 rather than a machine_mode.
10331 * expmed.c (mask_rtx): Likewise.
10332 (init_expmed_one_conv): Likewise.
10333 (expand_mult_highpart_adjust): Likewise.
10334 (extract_high_half): Likewise.
10335 (expmed_mult_highpart_optab): Likewise.
10336 (expmed_mult_highpart): Likewise.
10337 (expand_smod_pow2): Likewise.
10338 (expand_sdiv_pow2): Likewise.
10339 (emit_store_flag_int): Likewise.
10340 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10342 (extract_low_bits): Likewise.
10343 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10345 * expr.c (pieces_addr::adjust): Likewise.
10346 (can_store_by_pieces): Likewise.
10347 (store_by_pieces): Likewise.
10348 (clear_by_pieces_1): Likewise.
10349 (expand_expr_addr_expr_1): Likewise.
10350 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10351 (expand_expr_real_1): Likewise.
10352 (try_casesi): Likewise.
10353 * final.c (shorten_branches): Likewise.
10354 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10355 type of "mode" to machine_mode.
10356 * internal-fn.c (expand_arith_overflow_result_store): Take a
10357 scalar_int_mode rather than a machine_mode.
10358 (expand_mul_overflow): Use scalar_int_mode for local variables.
10359 * loop-doloop.c (doloop_modify): Likewise.
10360 (doloop_optimize): Likewise.
10361 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10362 than a machine_mode.
10363 (expand_doubleword_shift_condmove): Likewise.
10364 (expand_doubleword_shift): Likewise.
10365 (expand_doubleword_clz): Likewise.
10366 (expand_doubleword_popcount): Likewise.
10367 (expand_doubleword_parity): Likewise.
10368 (expand_absneg_bit): Use scalar_int_mode for local variables.
10369 (prepare_float_lib_cmp): Likewise.
10370 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10371 rather than a machine_mode.
10372 (convert_memory_address_addr_space): Likewise.
10373 (get_mode_bounds): Likewise.
10374 (get_address_mode): Return a scalar_int_mode rather than a
10376 * rtlanal.c (get_address_mode): Likewise.
10377 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10378 than a machine_mode.
10379 * targhooks.c (default_mode_rep_extended): Likewise.
10380 (default_valid_pointer_mode): Likewise.
10381 (default_addr_space_valid_pointer_mode): Likewise.
10382 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10383 than a machine_mode.
10384 (default_libgcc_cmp_return_mode): Likewise.
10385 (default_libgcc_shift_count_mode): Likewise.
10386 (default_unwind_word_mode): Likewise.
10387 (default_addr_space_pointer_mode): Likewise.
10388 (default_addr_space_address_mode): Likewise.
10389 * targhooks.h (default_eh_return_filter_mode): Likewise.
10390 (default_libgcc_cmp_return_mode): Likewise.
10391 (default_libgcc_shift_count_mode): Likewise.
10392 (default_unwind_word_mode): Likewise.
10393 (default_addr_space_pointer_mode): Likewise.
10394 (default_addr_space_address_mode): Likewise.
10395 (default_mode_rep_extended): Take a scalar_int_mode rather than
10397 (default_valid_pointer_mode): Likewise.
10398 (default_addr_space_valid_pointer_mode): Likewise.
10399 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10401 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10402 rather than a machine_mode.
10403 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10404 for local variables.
10405 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10406 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10407 than a machine_mode.
10409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10410 Alan Hayward <alan.hayward@arm.com>
10411 David Sherwood <david.sherwood@arm.com>
10413 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10414 the mode argument to scalar_int_mode.
10415 (do_jump_by_parts_zero_rtx): Likewise.
10416 (do_jump_by_parts_equality_rtx): Likewise.
10417 (do_jump_by_parts_greater): Take a mode argument.
10418 (do_jump_by_parts_equality): Likewise.
10419 (do_jump_1): Update calls accordingly.
10421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10422 Alan Hayward <alan.hayward@arm.com>
10423 David Sherwood <david.sherwood@arm.com>
10425 * is-a.h (safe_dyn_cast): New function.
10426 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10427 (jump_table_for_label): Likewise.
10428 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10429 instead of an rtx_insn *.
10430 (shorten_branches): Use dyn_cast instead of LABEL_P and
10431 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10432 rtx_jump_table_data::get_data_mode.
10433 (final_scan_insn): Likewise.
10435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10436 Alan Hayward <alan.hayward@arm.com>
10437 David Sherwood <david.sherwood@arm.com>
10439 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10440 trying to combine a full-register integer set with a subreg
10443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10444 Alan Hayward <alan.hayward@arm.com>
10445 David Sherwood <david.sherwood@arm.com>
10447 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10448 that is always either address_mode or pointer_mode.
10450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10451 Alan Hayward <alan.hayward@arm.com>
10452 David Sherwood <david.sherwood@arm.com>
10454 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10455 when the two are known to be equal.
10457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10458 Alan Hayward <alan.hayward@arm.com>
10459 David Sherwood <david.sherwood@arm.com>
10461 * simplify-rtx.c (simplify_const_unary_operation): Use
10462 is_a <scalar_int_mode> instead of checking for a nonzero
10463 precision. Forcibly convert op_mode to a scalar_int_mode
10464 in that case. More clearly differentiate the operand and
10465 result modes and use the former when deciding what the value
10466 of a count-bits operation should be. Use is_int_mode instead
10467 of checking for a MODE_INT. Remove redundant check for whether
10468 this mode has a zero precision.
10470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10471 Alan Hayward <alan.hayward@arm.com>
10472 David Sherwood <david.sherwood@arm.com>
10474 * optabs.c (widen_leading): Change the type of the mode argument
10475 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10476 (widen_bswap): Likewise.
10477 (expand_parity): Likewise.
10478 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10479 (expand_ffs): Likewise.
10480 (epand_unop): Check for scalar integer modes before calling the
10483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10484 Alan Hayward <alan.hayward@arm.com>
10485 David Sherwood <david.sherwood@arm.com>
10487 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10489 (expand_expr_real_1): Update call accordingly.
10491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10492 Alan Hayward <alan.hayward@arm.com>
10493 David Sherwood <david.sherwood@arm.com>
10495 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10496 value_mode arguments. Use scalar_int_mode internally.
10497 (store_bit_field_1): Rename the new integer mode from imode
10498 to op0_mode and use it instead of GET_MODE (op0). Update calls
10499 to store_split_bit_field, store_bit_field_using_insv and
10500 store_fixed_bit_field.
10501 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10502 Use scalar_int_mode internally. Use a bit count rather than a mode
10503 when calculating the largest bit size for get_best_mode.
10504 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10505 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10506 Remove assertion that OP0 has a scalar integer mode.
10507 (store_split_bit_field): Add op0_mode and value_mode arguments.
10508 Update calls to extract_fixed_bit_field.
10509 (extract_bit_field_using_extv): Add an op0_mode argument.
10510 Use scalar_int_mode internally.
10511 (extract_bit_field_1): Rename the new integer mode from imode to
10512 op0_mode and use it instead of GET_MODE (op0). Update calls to
10513 extract_split_bit_field, extract_bit_field_using_extv and
10514 extract_fixed_bit_field.
10515 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10516 to extract_split_bit_field and extract_fixed_bit_field_1.
10517 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10518 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10519 on the target mode.
10520 (extract_split_bit_field): Add an op0_mode argument. Update call
10521 to extract_fixed_bit_field.
10523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10524 Alan Hayward <alan.hayward@arm.com>
10525 David Sherwood <david.sherwood@arm.com>
10527 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10528 * explow.c (hard_function_value): Likewise.
10529 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10530 convert_to_mode call outside the loop.
10531 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10532 for the mode iterator. Require the mode specified by max_pieces
10534 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10536 (copy_blkmode_to_reg): Likewise.
10537 (set_storage_via_setmem): Likewise.
10538 * optabs.c (prepare_cmp_insn): Likewise.
10539 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10540 * stor-layout.c (finish_bitfield_representative): Likewise.
10542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10543 Alan Hayward <alan.hayward@arm.com>
10544 David Sherwood <david.sherwood@arm.com>
10546 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10547 * expr.c (convert_move): Use them.
10548 (convert_modes): Likewise.
10549 (store_expr_with_bounds): Likewise.
10551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10552 Alan Hayward <alan.hayward@arm.com>
10553 David Sherwood <david.sherwood@arm.com>
10555 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10556 parameter for the mode of "x". Remove the "known_x", "known_mode"
10557 and "known_ret" arguments. Change the type of the mode argument
10558 to scalar_int_mode.
10559 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10560 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10561 (reg_num_sign_bit_copies_for_combine): Likewise.
10562 * rtlanal.c (nonzero_bits1): Likewise.
10563 (num_sign_bit_copies1): Likewise.
10564 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10565 (reg_num_sign_bit_copies_general): Likewise.
10566 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10567 (reg_nonzero_bits_general): Likewise.
10569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10570 Alan Hayward <alan.hayward@arm.com>
10571 David Sherwood <david.sherwood@arm.com>
10573 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10574 than in subroutines. Return 1 for non-integer modes.
10575 (cached_num_sign_bit_copies): Change the type of the mode parameter
10576 to scalar_int_mode.
10577 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10578 classes. Handle CONST_INT_P first and then check whether X also
10579 has a scalar integer mode. Check the same thing for inner registers
10580 of a SUBREG and for values that are being extended or truncated.
10582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10583 Alan Hayward <alan.hayward@arm.com>
10584 David Sherwood <david.sherwood@arm.com>
10586 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10587 in subroutines. Return the mode mask for non-integer modes.
10588 (cached_nonzero_bits): Change the type of the mode parameter
10589 to scalar_int_mode.
10590 (nonzero_bits1): Likewise. Remove early exit for other mode
10591 classes. Handle CONST_INT_P first and then check whether X
10592 also has a scalar integer mode.
10594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10595 Alan Hayward <alan.hayward@arm.com>
10596 David Sherwood <david.sherwood@arm.com>
10598 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10599 Return a scalar_int_mode and assert that the size is greater than
10601 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10602 (op_by_pieces_d::op_by_pieces_d): Likewise.
10603 (op_by_pieces_d::run): Likewise.
10604 (can_store_by_pieces): Likewise.
10606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10607 Alan Hayward <alan.hayward@arm.com>
10608 David Sherwood <david.sherwood@arm.com>
10610 * combine.c (extract_left_shift): Add a mode argument and update
10612 (make_compound_operation_int): Change the type of the mode parameter
10613 to scalar_int_mode and update the call to extract_left_shift.
10615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10616 Alan Hayward <alan.hayward@arm.com>
10617 David Sherwood <david.sherwood@arm.com>
10619 * combine.c (simplify_and_const_int): Change the type of the mode
10620 parameter to scalar_int_mode.
10621 (simplify_and_const_int_1): Likewise. Update recursive call.
10623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10624 Alan Hayward <alan.hayward@arm.com>
10625 David Sherwood <david.sherwood@arm.com>
10627 * combine.c (simplify_compare_const): Check that the mode is a
10628 scalar_int_mode (rather than VOIDmode) before testing its
10630 (simplify_comparison): Move COMPARISON_P handling out of the
10631 loop and restrict the latter part of the loop to scalar_int_modes.
10632 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10633 and when considering SUBREG_REGs. Use is_int_mode instead of
10634 checking GET_MODE_CLASS against MODE_INT.
10636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10637 Alan Hayward <alan.hayward@arm.com>
10638 David Sherwood <david.sherwood@arm.com>
10640 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10641 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10642 shift_unit_mode and for modes involved in scalar shifts.
10644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10645 Alan Hayward <alan.hayward@arm.com>
10646 David Sherwood <david.sherwood@arm.com>
10648 * combine.c (force_int_to_mode): New function, split out from...
10649 (force_to_mode): ...here. Keep xmode up-to-date and use it
10650 instead of GET_MODE (x).
10652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10653 Alan Hayward <alan.hayward@arm.com>
10654 David Sherwood <david.sherwood@arm.com>
10656 * optabs-query.h (extraction_insn::struct_mode): Change type to
10657 opt_scalar_int_mode and update comment.
10658 (extraction_insn::field_mode): Change type to scalar_int_mode.
10659 (extraction_insn::pos_mode): Likewise.
10660 * combine.c (make_extraction): Update accordingly.
10661 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10662 (get_optab_extraction_insn): Likewise.
10663 * recog.c (simplify_while_replacing): Likewise.
10664 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10665 parameter to opt_scalar_int_mode.
10667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10668 Alan Hayward <alan.hayward@arm.com>
10669 David Sherwood <david.sherwood@arm.com>
10671 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10672 to a scalar_int_mode instead of a machine_mode.
10673 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10674 (get_best_mode): Return a boolean and use a pointer argument to store
10675 the selected mode. Replace the limit mode parameter with a bit limit.
10676 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10677 for the values returned by bit_field_mode_iterator::next_mode.
10678 (store_bit_field): Update call to get_best_mode.
10679 (store_fixed_bit_field): Likewise.
10680 (extract_fixed_bit_field): Likewise.
10681 * expr.c (optimize_bitfield_assignment_op): Likewise.
10682 * fold-const.c (optimize_bit_field_compare): Likewise.
10683 (fold_truth_andor_1): Likewise.
10684 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10685 Update for new type of m_mode.
10686 (get_best_mode): As above.
10688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10689 Alan Hayward <alan.hayward@arm.com>
10690 David Sherwood <david.sherwood@arm.com>
10692 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10693 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10694 (store_bit_field): Check is_a <scalar_int_mode> before calling
10695 strict_volatile_bitfield_p.
10696 (extract_bit_field): Likewise.
10698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10699 Alan Hayward <alan.hayward@arm.com>
10700 David Sherwood <david.sherwood@arm.com>
10702 * target.def (cstore_mode): Return a scalar_int_mode.
10703 * doc/tm.texi: Regenerate.
10704 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10705 * targhooks.h (default_cstore_mode): Likewise.
10706 * targhooks.c (default_cstore_mode): Likewise, using a forced
10708 * expmed.c (emit_cstore): Expect the target of the cstore to be
10711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10712 Alan Hayward <alan.hayward@arm.com>
10713 David Sherwood <david.sherwood@arm.com>
10715 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10717 (niter_desc): Likewise mode.
10718 (iv_analyze): Add a mode parameter.
10720 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10721 and change its type to scalar_int_mode.
10722 * loop-iv.c: Update commentary at head of file.
10723 (iv_constant): Pass the mode paraeter before the rtx it describes
10724 and change its type to scalar_int_mode. Remove VOIDmode handling.
10725 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10726 (iv_extend): Likewise.
10727 (shorten_into_mode): Likewise.
10728 (iv_add): Use scalar_int_mode.
10729 (iv_mult): Likewise.
10730 (iv_shift): Likewise.
10731 (canonicalize_iv_subregs): Likewise.
10732 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10733 it describes and change its mode to scalar_int_mode. Also change
10734 the type of the returned inner_mode to scalar_int_mode.
10735 (get_biv_step): Likewise, turning outer_mode from a pointer
10736 into a direct parameter. Update call to get_biv_step_1.
10737 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10738 iv_constant and get_biv_step.
10739 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10740 and change its type to scalar_int_mode. Don't initialise iv->mode
10741 to VOIDmode and remove later checks for its still being VOIDmode.
10742 Update calls to iv_analyze_op and iv_analyze_expr. Check
10743 is_a <scalar_int_mode> when changing the mode under consideration.
10744 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10745 Update call to iv_analyze_expr.
10746 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10747 inner register is not also a scalar_int_mode. Update call to
10749 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10750 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10751 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10752 separate mode class checks. Update calls to iv_analyze. Remove
10753 fix-up of VOIDmodes after iv_analyze_biv.
10754 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10755 don't have a scalar_int_mode. Update call to biv_p.
10757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10758 Alan Hayward <alan.hayward@arm.com>
10759 David Sherwood <david.sherwood@arm.com>
10761 * cfgexpand.c (convert_debug_memory_address): Use
10762 as_a <scalar_int_mode>.
10763 * combine.c (expand_compound_operation): Likewise.
10764 (make_extraction): Likewise.
10765 (change_zero_ext): Likewise.
10766 (simplify_comparison): Likewise.
10767 * cse.c (cse_insn): Likewise.
10768 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10769 (mem_loc_descriptor): Likewise.
10770 (loc_descriptor): Likewise.
10771 * expmed.c (init_expmed_one_mode): Likewise.
10772 (synth_mult): Likewise.
10773 (emit_store_flag_1): Likewise.
10774 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10775 of a comparison with size.
10776 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10777 (reduce_to_bit_field_precision): Likewise.
10778 * function.c (expand_function_end): Likewise.
10779 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10780 * loop-doloop.c (doloop_modify): Likewise.
10781 * optabs.c (expand_binop): Likewise.
10782 (expand_unop): Likewise.
10783 (expand_copysign_absneg): Likewise.
10784 (prepare_cmp_insn): Likewise.
10785 (maybe_legitimize_operand): Likewise.
10786 * recog.c (const_scalar_int_operand): Likewise.
10787 * rtlanal.c (get_address_mode): Likewise.
10788 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10789 (simplify_cond_clz_ctz): Likewise.
10790 * tree-nested.c (get_nl_goto_field): Likewise.
10791 * tree.c (build_vector_type_for_mode): Likewise.
10792 * var-tracking.c (use_narrower_mode): Likewise.
10794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10795 Alan Hayward <alan.hayward@arm.com>
10796 David Sherwood <david.sherwood@arm.com>
10798 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10799 * builtins.c (expand_builtin_signbit): Use it.
10800 * cfgexpand.c (expand_debug_expr): Likewise.
10801 * dojump.c (do_jump): Likewise.
10802 (do_compare_and_jump): Likewise.
10803 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10804 * expmed.c (make_tree): Likewise.
10805 * expr.c (expand_expr_real_2): Likewise.
10806 (expand_expr_real_1): Likewise.
10807 (try_casesi): Likewise.
10808 * fold-const-call.c (fold_const_call_ss): Likewise.
10809 * fold-const.c (unextend): Likewise.
10810 (extract_muldiv_1): Likewise.
10811 (fold_single_bit_test): Likewise.
10812 (native_encode_int): Likewise.
10813 (native_encode_string): Likewise.
10814 (native_interpret_int): Likewise.
10815 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10816 * internal-fn.c (expand_addsub_overflow): Likewise.
10817 (expand_neg_overflow): Likewise.
10818 (expand_mul_overflow): Likewise.
10819 (expand_arith_overflow): Likewise.
10820 * match.pd: Likewise.
10821 * stor-layout.c (layout_type): Likewise.
10822 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10823 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10824 * tree-ssanames.c (get_range_info): Likewise.
10825 * tree-switch-conversion.c (array_value_type) Likewise.
10826 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10827 (vect_recog_divmod_pattern): Likewise.
10828 (vect_recog_mixed_size_cond_pattern): Likewise.
10829 * tree-vrp.c (extract_range_basic): Likewise.
10830 (simplify_float_conversion_using_ranges): Likewise.
10831 * tree.c (int_fits_type_p): Likewise.
10832 * ubsan.c (instrument_bool_enum_load): Likewise.
10833 * varasm.c (mergeable_string_section): Likewise.
10834 (narrowing_initializer_constant_valid_p): Likewise.
10835 (output_constant): Likewise.
10837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10838 Alan Hayward <alan.hayward@arm.com>
10839 David Sherwood <david.sherwood@arm.com>
10841 * machmode.h (NARROWEST_INT_MODE): New macro.
10842 * expr.c (alignment_for_piecewise_move): Use it instead of
10843 GET_CLASS_NARROWEST_MODE (MODE_INT).
10844 (push_block): Likewise.
10845 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10847 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10850 Alan Hayward <alan.hayward@arm.com>
10851 David Sherwood <david.sherwood@arm.com>
10853 * postreload.c (move2add_valid_value_p): Change the type of the
10854 mode parameter to scalar_int_mode.
10855 (move2add_use_add2_insn): Add a mode parameter and use it instead
10857 (move2add_use_add3_insn): Likewise.
10858 (reload_cse_move2add): Update accordingly.
10860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10861 Alan Hayward <alan.hayward@arm.com>
10862 David Sherwood <david.sherwood@arm.com>
10864 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10866 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10867 * optabs.c (expand_unop): Likewise.
10869 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10870 Alan Hayward <alan.hayward@arm.com>
10871 David Sherwood <david.sherwood@arm.com>
10873 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10874 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10875 (popcount_loc_descriptor): Likewise.
10876 (bswap_loc_descriptor): Likewise.
10877 (rotate_loc_descriptor): Likewise.
10878 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10879 calling the functions above.
10881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10882 Alan Hayward <alan.hayward@arm.com>
10883 David Sherwood <david.sherwood@arm.com>
10885 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10887 (try_combine): Likewise.
10888 (simplify_if_then_else): Likewise.
10889 * cse.c (cse_insn): Likewise.
10890 * dwarf2out.c (mem_loc_descriptor): Likewise.
10891 * emit-rtl.c (gen_lowpart_common): Likewise.
10892 * simplify-rtx.c (simplify_truncation): Likewise.
10893 (simplify_binary_operation_1): Likewise.
10894 (simplify_const_relational_operation): Likewise.
10895 (simplify_ternary_operation): Likewise.
10896 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10899 Alan Hayward <alan.hayward@arm.com>
10900 David Sherwood <david.sherwood@arm.com>
10902 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10903 * reload.c (push_reload): Likewise.
10904 (find_reloads): Likewise.
10906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10907 Alan Hayward <alan.hayward@arm.com>
10908 David Sherwood <david.sherwood@arm.com>
10910 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10911 (make_compound_operation_int): Likewise.
10912 (change_zero_ext): Likewise.
10913 * expr.c (convert_move): Likewise.
10914 (convert_modes): Likewise.
10915 * fwprop.c (forward_propagate_subreg): Likewise.
10916 * loop-iv.c (get_biv_step_1): Likewise.
10917 * optabs.c (widen_operand): Likewise.
10918 * postreload.c (move2add_valid_value_p): Likewise.
10919 * recog.c (simplify_while_replacing): Likewise.
10920 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10921 (simplify_binary_operation_1): Likewise. Remove redundant
10922 mode equality check.
10924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10925 Alan Hayward <alan.hayward@arm.com>
10926 David Sherwood <david.sherwood@arm.com>
10928 * combine.c (combine_simplify_rtx): Add checks for
10929 is_a <scalar_int_mode>.
10930 (simplify_if_then_else): Likewise.
10931 (make_field_assignment): Likewise.
10932 (simplify_comparison): Likewise.
10933 * ifcvt.c (noce_try_bitop): Likewise.
10934 * loop-invariant.c (canonicalize_address_mult): Likewise.
10935 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10938 Alan Hayward <alan.hayward@arm.com>
10939 David Sherwood <david.sherwood@arm.com>
10941 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10942 is_a <scalar_int_mode> instead of != BLKmode.
10944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10945 Alan Hayward <alan.hayward@arm.com>
10946 David Sherwood <david.sherwood@arm.com>
10948 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10949 instead of != VOIDmode.
10950 * combine.c (if_then_else_cond): Likewise.
10951 (change_zero_ext): Likewise.
10952 * dwarf2out.c (mem_loc_descriptor): Likewise.
10953 (loc_descriptor): Likewise.
10954 * rtlanal.c (canonicalize_condition): Likewise.
10955 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10958 Alan Hayward <alan.hayward@arm.com>
10959 David Sherwood <david.sherwood@arm.com>
10961 * simplify-rtx.c (simplify_binary_operation_1): Use
10962 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10965 Alan Hayward <alan.hayward@arm.com>
10966 David Sherwood <david.sherwood@arm.com>
10968 * wide-int.h (int_traits<unsigned char>) New class.
10969 (int_traits<unsigned short>) Likewise.
10970 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10971 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10973 * combine.c (set_nonzero_bits_and_sign_copies): Use
10974 is_a <scalar_int_mode>.
10975 (find_split_point): Likewise.
10976 (combine_simplify_rtx): Likewise.
10977 (simplify_logical): Likewise.
10978 (expand_compound_operation): Likewise.
10979 (expand_field_assignment): Likewise.
10980 (make_compound_operation): Likewise.
10981 (extended_count): Likewise.
10982 (change_zero_ext): Likewise.
10983 (simplify_comparison): Likewise.
10984 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10985 (ucompare_loc_descriptor): Likewise.
10986 (minmax_loc_descriptor): Likewise.
10987 (mem_loc_descriptor): Likewise.
10988 (loc_descriptor): Likewise.
10989 * expmed.c (init_expmed_one_mode): Likewise.
10990 * lra-constraints.c (lra_constraint_offset): Likewise.
10991 * optabs.c (prepare_libcall_arg): Likewise.
10992 * postreload.c (move2add_note_store): Likewise.
10993 * reload.c (operands_match_p): Likewise.
10994 * rtl.h (load_extend_op): Likewise.
10995 * rtlhooks.c (gen_lowpart_general): Likewise.
10996 * simplify-rtx.c (simplify_truncation): Likewise.
10997 (simplify_unary_operation_1): Likewise.
10998 (simplify_binary_operation_1): Likewise.
10999 (simplify_const_binary_operation): Likewise.
11000 (simplify_const_relational_operation): Likewise.
11001 (simplify_subreg): Likewise.
11002 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11003 * var-tracking.c (adjust_mems): Likewise.
11004 (prepare_call_arguments): Likewise.
11006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11007 Alan Hayward <alan.hayward@arm.com>
11008 David Sherwood <david.sherwood@arm.com>
11010 * machmode.h (is_int_mode): New fuction.
11011 * combine.c (find_split_point): Use it.
11012 (combine_simplify_rtx): Likewise.
11013 (simplify_if_then_else): Likewise.
11014 (simplify_set): Likewise.
11015 (simplify_shift_const_1): Likewise.
11016 (simplify_comparison): Likewise.
11017 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11018 * cse.c (notreg_cost): Likewise.
11019 (cse_insn): Likewise.
11020 * cselib.c (cselib_lookup_1): Likewise.
11021 * dojump.c (do_jump_1): Likewise.
11022 (do_compare_rtx_and_jump): Likewise.
11023 * dse.c (get_call_args): Likewise.
11024 * dwarf2out.c (rtl_for_decl_init): Likewise.
11025 (native_encode_initializer): Likewise.
11026 * expmed.c (emit_store_flag_1): Likewise.
11027 (emit_store_flag): Likewise.
11028 * expr.c (convert_modes): Likewise.
11029 (store_field): Likewise.
11030 (expand_expr_real_1): Likewise.
11031 * fold-const.c (fold_read_from_constant_string): Likewise.
11032 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11033 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11034 * optabs.c (expand_binop): Likewise.
11035 (expand_unop): Likewise.
11036 (expand_abs_nojump): Likewise.
11037 (expand_one_cmpl_abs_nojump): Likewise.
11038 * simplify-rtx.c (mode_signbit_p): Likewise.
11039 (val_signbit_p): Likewise.
11040 (val_signbit_known_set_p): Likewise.
11041 (val_signbit_known_clear_p): Likewise.
11042 (simplify_relational_operation_1): Likewise.
11043 * tree.c (vector_type_mode): Likewise.
11045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11046 Alan Hayward <alan.hayward@arm.com>
11047 David Sherwood <david.sherwood@arm.com>
11049 * machmode.h (smallest_mode_for_size): Fix formatting.
11050 (smallest_int_mode_for_size): New function.
11051 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11052 instead of smallest_mode_for_size.
11053 * combine.c (make_extraction): Likewise.
11054 * config/arc/arc.c (arc_expand_movmem): Likewise.
11055 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11056 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11057 * config/s390/s390.c (s390_expand_insv): Likewise.
11058 * config/sparc/sparc.c (assign_int_registers): Likewise.
11059 * config/spu/spu.c (spu_function_value): Likewise.
11060 (spu_function_arg): Likewise.
11061 * coverage.c (get_gcov_type): Likewise.
11062 (get_gcov_unsigned_t): Likewise.
11063 * dse.c (find_shift_sequence): Likewise.
11064 * expmed.c (store_bit_field_1): Likewise.
11065 * expr.c (convert_move): Likewise.
11066 (store_field): Likewise.
11067 * internal-fn.c (expand_arith_overflow): Likewise.
11068 * optabs-query.c (get_best_extraction_insn): Likewise.
11069 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11070 * stor-layout.c (layout_type): Likewise.
11071 (initialize_sizetypes): Likewise.
11072 * targhooks.c (default_get_mask_mode): Likewise.
11073 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11076 Alan Hayward <alan.hayward@arm.com>
11077 David Sherwood <david.sherwood@arm.com>
11079 * machmode.h (opt_mode::else_blk): New function.
11080 (int_mode_for_mode): Declare.
11081 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11082 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11084 * cfgexpand.c (expand_debug_expr): Likewise.
11085 * combine.c (gen_lowpart_or_truncate): Likewise.
11086 (gen_lowpart_for_combine): Likewise.
11087 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11088 * config/avr/avr.c (avr_to_int_mode): Likewise.
11089 (avr_out_plus_1): Likewise.
11090 (avr_out_plus): Likewise.
11091 (avr_out_round): Likewise.
11092 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11093 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11094 (s390_expand_vcond): Likewise.
11095 * config/spu/spu.c (spu_split_immediate): Likewise.
11096 (spu_expand_mov): Likewise.
11097 * dse.c (get_stored_val): Likewise.
11098 * expmed.c (store_bit_field_1): Likewise.
11099 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11101 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11102 (extract_low_bits): Likewise.
11103 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11104 handling rather than repeating the check.
11105 (emit_group_store): Likewise.
11106 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11107 * optabs.c (expand_absneg_bit): Likewise.
11108 (expand_copysign_absneg): Likewise.
11109 (expand_copysign_bit): Likewise.
11110 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11111 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11112 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11113 * var-tracking.c (prepare_call_arguments): Likewise.
11114 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11115 int_mode_for_mode instead of mode_for_size.
11116 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11119 Alan Hayward <alan.hayward@arm.com>
11120 David Sherwood <david.sherwood@arm.com>
11122 * machmode.h (int_mode_for_size): New function.
11123 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11124 instead of mode_for_size.
11125 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11127 * combine.c (expand_field_assignment): Use int_mode_for_size
11128 instead of mode_for_size.
11129 (make_extraction): Likewise.
11130 (simplify_shift_const_1): Likewise.
11131 (simplify_comparison): Likewise.
11132 * dojump.c (do_jump): Likewise.
11133 * dwarf2out.c (mem_loc_descriptor): Likewise.
11134 * emit-rtl.c (init_derived_machine_modes): Likewise.
11135 * expmed.c (flip_storage_order): Likewise.
11136 (convert_extracted_bit_field): Likewise.
11137 * expr.c (copy_blkmode_from_reg): Likewise.
11138 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11139 * internal-fn.c (expand_mul_overflow): Likewise.
11140 * lower-subreg.c (simple_move): Likewise.
11141 * optabs-libfuncs.c (init_optabs): Likewise.
11142 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11143 * tree.c (vector_type_mode): Likewise.
11144 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11145 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11146 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11147 * tree-vect-stmts.c (vectorizable_load): Likewise.
11148 (vectorizable_store): Likewise.
11150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11151 Alan Hayward <alan.hayward@arm.com>
11152 David Sherwood <david.sherwood@arm.com>
11154 * coretypes.h (pod_mode): New type.
11155 (scalar_int_mode_pod): New typedef.
11156 * machmode.h (pod_mode): New class.
11157 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11158 * genmodes.c (emit_mode_int_n): Update accordingly.
11159 * lower-subreg.h (target_lower_subreg): Change type to
11160 scalar_int_mode_pod.
11161 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11162 scalar_int_mode_pod.
11164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11165 Alan Hayward <alan.hayward@arm.com>
11166 David Sherwood <david.sherwood@arm.com>
11168 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11169 machine_mode to scalar_int_mode.
11170 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11171 (rs6000_option_override_internal): Remove cast to int.
11172 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11173 machine_mode to scalar_int_mode.
11174 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11175 (rs6000_option_override_internal): Remove cast to int.
11176 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11177 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11179 * config/s390/s390.c (s390_expand_builtin): Likewise.
11180 * coretypes.h (scalar_int_mode): New type.
11181 (opt_scalar_int_mode): New typedef.
11182 * machmode.h (scalar_int_mode): New class.
11183 (scalar_int_mode::includes_p): New function.
11184 (byte_mode): Change type to scalar_int_mode.
11185 (word_mode): Likewise.
11186 (ptr_mode): Likewise.
11187 * emit-rtl.c (byte_mode): Likewise.
11188 (word_mode): Likewise.
11189 (ptr_mode): Likewise.
11190 (init_derived_machine_modes): Update accordingly.
11191 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11192 and MODE_PARTIAL_INT.
11193 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11194 opt_scalar_int_mode.
11196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11197 Alan Hayward <alan.hayward@arm.com>
11198 David Sherwood <david.sherwood@arm.com>
11200 * target.def (libgcc_floating_mode_supported_p): Take a
11202 * doc/tm.texi: Regenerate.
11203 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11205 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11206 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11210 Alan Hayward <alan.hayward@arm.com>
11211 David Sherwood <david.sherwood@arm.com>
11213 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11214 * doc/tm.texi: Regenerate.
11215 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11216 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11217 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11218 * targhooks.h (default_floatn_mode): Likewise.
11219 * targhooks.c (default_floatn_mode): Likewise.
11220 * tree.c (build_common_tree_nodes): Update accordingly.
11222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11223 Alan Hayward <alan.hayward@arm.com>
11224 David Sherwood <david.sherwood@arm.com>
11226 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11227 (mode_iterator::iterate_p): Likewise.
11228 (mode_iterator::get_wider): Likewise.
11229 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11232 Alan Hayward <alan.hayward@arm.com>
11233 David Sherwood <david.sherwood@arm.com>
11235 * coretypes.h (opt_scalar_float_mode): New typedef.
11236 * machmode.h (float_mode_for_size): New function.
11237 * emit-rtl.c (double_mode): Delete.
11238 (init_emit_once): Use float_mode_for_size.
11239 * stor-layout.c (layout_type): Likewise.
11240 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11243 Alan Hayward <alan.hayward@arm.com>
11244 David Sherwood <david.sherwood@arm.com>
11246 * output.h (assemble_real): Take a scalar_float_mode.
11247 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11248 * config/arm/arm.md (consttable_4): Likewise.
11249 (consttable_8): Likewise.
11250 (consttable_16): Likewise.
11251 * config/mips/mips.md (consttable_float): Likewise.
11252 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11253 * varasm.c (assemble_real): Take a scalar_float_mode.
11254 (output_constant_pool_2): Update accordingly.
11255 (output_constant): Likewise.
11257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11258 Alan Hayward <alan.hayward@arm.com>
11259 David Sherwood <david.sherwood@arm.com>
11261 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11262 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11263 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11264 (native_encode_real): Likewise.
11265 (native_interpret_real): Likewise.
11266 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11267 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11270 Alan Hayward <alan.hayward@arm.com>
11271 David Sherwood <david.sherwood@arm.com>
11273 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11274 <scalar_float_mode>. Simplify.
11275 (gen_extend_conv_libfunc): Likewise.
11277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11278 Alan Hayward <alan.hayward@arm.com>
11279 David Sherwood <david.sherwood@arm.com>
11281 * coretypes.h (scalar_float_mode): New type.
11282 * machmode.h (mode_traits::from_int): Use machine_mode if
11283 USE_ENUM_MODES is defined.
11284 (is_a): New function.
11286 (dyn_cast): Likewise.
11287 (scalar_float_mode): New class.
11288 (scalar_float_mode::includes_p): New function.
11289 (is_float_mode): Likewise.
11290 * gdbhooks.py (MachineModePrinter): New class.
11291 (build_pretty_printer): Use it for scalar_float_mode.
11292 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11293 (format_helper::format_helper): Turn into a template.
11294 * genmodes.c (get_mode_class): New function.
11295 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11296 or machine_mode if none.
11297 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11298 as_a <scalar_float_mode>.
11299 * dwarf2out.c (mem_loc_descriptor): Likewise.
11300 (insert_float): Likewise.
11301 (add_const_value_attribute): Likewise.
11302 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11303 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11304 (expand_unop): Update accordingly.
11305 (expand_abs_nojump): Likewise.
11306 (expand_copysign_absneg): Take a scalar_float_mode.
11307 (expand_copysign_bit): Likewise.
11308 (expand_copysign): Update accordingly.
11310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11311 Alan Hayward <alan.hayward@arm.com>
11312 David Sherwood <david.sherwood@arm.com>
11314 * coretypes.h (opt_mode): New class.
11315 * machmode.h (opt_mode): Likewise.
11316 (opt_mode::else_void): New function.
11317 (opt_mode::require): Likewise.
11318 (opt_mode::exists): Likewise.
11319 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11320 (GET_MODE_2XWIDER_MODE): Likewise.
11321 (mode_iterator::get_wider): Update accordingly.
11322 (mode_iterator::get_2xwider): Likewise.
11323 (mode_iterator::get_known_wider): Likewise, turning into a template.
11324 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11325 forcing a wider mode to exist.
11326 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11327 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11328 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11329 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11330 * lower-subreg.c (init_lower_subreg): Likewise.
11331 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11332 on the final iteration.
11333 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11334 a wider mode exists before asking for a move pattern.
11335 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11336 forcing a wider mode to exist.
11337 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11338 returning false if no such mode exists.
11339 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11340 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11341 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11342 Avoid checking for a MODE_INT if we already know the mode is not a
11344 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11345 forcing a wider mode to exist.
11346 (expmed_mult_highpart_optab): Likewise.
11347 (expmed_mult_highpart): Likewise.
11348 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11350 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11351 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11352 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11353 * internal-fn.c (expand_mul_overflow): Update use of
11354 GET_MODE_2XWIDER_MODE.
11355 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11356 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11357 GET_MODE_WIDER_MODE.
11358 (convert_plusminus_to_widen): Likewise.
11359 * tree-switch-conversion.c (array_value_type): Likewise.
11360 * var-tracking.c (emit_note_insn_var_location): Likewise.
11361 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11362 Return false inside rather than outside the loop if no wider mode
11364 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11365 and GET_MODE_2XWIDER_MODE
11366 (can_compare_p): Use else_void.
11367 * gdbhooks.py (OptMachineModePrinter): New class.
11368 (build_pretty_printer): Use it for opt_mode.
11370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11371 Alan Hayward <alan.hayward@arm.com>
11372 David Sherwood <david.sherwood@arm.com>
11374 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11375 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11378 Alan Hayward <alan.hayward@arm.com>
11379 David Sherwood <david.sherwood@arm.com>
11381 * machmode.h (mode_traits): New structure.
11382 (get_narrowest_mode): New function.
11383 (mode_iterator::start): Likewise.
11384 (mode_iterator::iterate_p): Likewise.
11385 (mode_iterator::get_wider): Likewise.
11386 (mode_iterator::get_known_wider): Likewise.
11387 (mode_iterator::get_2xwider): Likewise.
11388 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11389 (FOR_EACH_MODE): Likewise.
11390 (FOR_EACH_MODE_FROM): Likewise.
11391 (FOR_EACH_MODE_UNTIL): Likewise.
11392 (FOR_EACH_WIDER_MODE): Likewise.
11393 (FOR_EACH_2XWIDER_MODE): Likewise.
11394 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11395 * combine.c (simplify_comparison): Likewise
11396 * config/i386/i386.c (type_natural_mode): Likewise.
11397 * cse.c (cse_insn): Likewise.
11398 * dse.c (find_shift_sequence): Likewise.
11399 * emit-rtl.c (init_derived_machine_modes): Likewise.
11400 (init_emit_once): Likewise.
11401 * explow.c (hard_function_value): Likewise.
11402 * expmed.c (extract_fixed_bit_field_1): Likewise.
11403 (extract_bit_field_1): Likewise.
11404 (expand_divmod): Likewise.
11405 (emit_store_flag_1): Likewise.
11406 * expr.c (init_expr_target): Likewise.
11407 (convert_move): Likewise.
11408 (alignment_for_piecewise_move): Likewise.
11409 (widest_int_mode_for_size): Likewise.
11410 (emit_block_move_via_movmem): Likewise.
11411 (copy_blkmode_to_reg): Likewise.
11412 (set_storage_via_setmem): Likewise.
11413 (compress_float_constant): Likewise.
11414 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11415 * optabs-query.c (get_best_extraction_insn): Likewise.
11416 * optabs.c (expand_binop): Likewise.
11417 (expand_twoval_unop): Likewise.
11418 (expand_twoval_binop): Likewise.
11419 (widen_leading): Likewise.
11420 (widen_bswap): Likewise.
11421 (expand_parity): Likewise.
11422 (expand_unop): Likewise.
11423 (prepare_cmp_insn): Likewise.
11424 (prepare_float_lib_cmp): Likewise.
11425 (expand_float): Likewise.
11426 (expand_fix): Likewise.
11427 (expand_sfix_optab): Likewise.
11428 * postreload.c (move2add_use_add2_insn): Likewise.
11429 * reg-stack.c (reg_to_stack): Likewise.
11430 * reginfo.c (choose_hard_reg_mode): Likewise.
11431 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11432 * stor-layout.c (mode_for_size): Likewise.
11433 (smallest_mode_for_size): Likewise.
11434 (mode_for_vector): Likewise.
11435 (finish_bitfield_representative): Likewise.
11436 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11437 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11438 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11439 * var-tracking.c (prepare_call_arguments): Likewise.
11441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11442 Alan Hayward <alan.hayward@arm.com>
11443 David Sherwood <david.sherwood@arm.com>
11445 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11446 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11447 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11448 * machmode.h (mode_size): Move earlier in file.
11449 (mode_precision): Likewise.
11450 (mode_inner): Likewise.
11451 (mode_nunits): Likewise.
11452 (mode_unit_size): Likewise.
11453 (unit_unit_precision): Likewise.
11454 (mode_wider): Likewise.
11455 (mode_2xwider): Likewise.
11456 (machine_mode): New class.
11457 (mode_to_bytes): New function.
11458 (mode_to_bits): Likewise.
11459 (mode_to_precision): Likewise.
11460 (mode_to_inner): Likewise.
11461 (mode_to_unit_size): Likewise.
11462 (mode_to_unit_precision): Likewise.
11463 (mode_to_nunits): Likewise.
11464 (GET_MODE_SIZE): Use mode_to_bytes.
11465 (GET_MODE_BITSIZE): Use mode_to_bits.
11466 (GET_MODE_PRECISION): Use mode_to_precision.
11467 (GET_MODE_INNER): Use mode_to_inner.
11468 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11469 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11470 (GET_MODE_NUNITS): Use mode_to_nunits.
11471 * system.h (ALWAYS_INLINE): New macro.
11472 * config/powerpcspe/powerpcspe-c.c
11473 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11474 int for arg1_mode and arg2_mode.
11476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11477 Alan Hayward <alan.hayward@arm.com>
11478 David Sherwood <david.sherwood@arm.com>
11480 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11481 Prefix mode names with E_ in case statements.
11482 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11483 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11484 (aarch64_split_simd_move): Likewise.
11485 (aarch64_gen_storewb_pair): Likewise.
11486 (aarch64_gen_loadwb_pair): Likewise.
11487 (aarch64_gen_store_pair): Likewise.
11488 (aarch64_gen_load_pair): Likewise.
11489 (aarch64_get_condition_code_1): Likewise.
11490 (aarch64_constant_pool_reload_icode): Likewise.
11491 (get_rsqrte_type): Likewise.
11492 (get_rsqrts_type): Likewise.
11493 (get_recpe_type): Likewise.
11494 (get_recps_type): Likewise.
11495 (aarch64_gimplify_va_arg_expr): Likewise.
11496 (aarch64_simd_container_mode): Likewise.
11497 (aarch64_emit_load_exclusive): Likewise.
11498 (aarch64_emit_store_exclusive): Likewise.
11499 (aarch64_expand_compare_and_swap): Likewise.
11500 (aarch64_gen_atomic_cas): Likewise.
11501 (aarch64_emit_bic): Likewise.
11502 (aarch64_emit_atomic_swap): Likewise.
11503 (aarch64_emit_atomic_load_op): Likewise.
11504 (aarch64_evpc_trn): Likewise.
11505 (aarch64_evpc_uzp): Likewise.
11506 (aarch64_evpc_zip): Likewise.
11507 (aarch64_evpc_ext): Likewise.
11508 (aarch64_evpc_rev): Likewise.
11509 (aarch64_evpc_dup): Likewise.
11510 (aarch64_gen_ccmp_first): Likewise.
11511 (aarch64_gen_ccmp_next): Likewise.
11512 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11513 (alpha_emit_xfloating_libcall): Likewise.
11514 (emit_insxl): Likewise.
11515 (alpha_arg_type): Likewise.
11516 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11517 (arc_preferred_simd_mode): Likewise.
11518 (arc_secondary_reload): Likewise.
11519 (get_arc_condition_code): Likewise.
11520 (arc_print_operand): Likewise.
11521 (arc_legitimate_constant_p): Likewise.
11522 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11523 * config/arc/arc.md (casesi_load): Likewise.
11524 (casesi_compact_jump): Likewise.
11525 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11526 (cc_use_register): Likewise.
11527 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11528 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11529 (arm_init_iwmmxt_builtins): Likewise.
11530 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11531 (neon_expand_vector_init): Likewise.
11532 (arm_attr_length_move_neon): Likewise.
11533 (maybe_get_arm_condition_code): Likewise.
11534 (arm_emit_vector_const): Likewise.
11535 (arm_preferred_simd_mode): Likewise.
11536 (arm_output_iwmmxt_tinsr): Likewise.
11537 (thumb1_output_casesi): Likewise.
11538 (thumb2_output_casesi): Likewise.
11539 (arm_emit_load_exclusive): Likewise.
11540 (arm_emit_store_exclusive): Likewise.
11541 (arm_expand_compare_and_swap): Likewise.
11542 (arm_evpc_neon_vuzp): Likewise.
11543 (arm_evpc_neon_vzip): Likewise.
11544 (arm_evpc_neon_vrev): Likewise.
11545 (arm_evpc_neon_vtrn): Likewise.
11546 (arm_evpc_neon_vext): Likewise.
11547 (arm_validize_comparison): Likewise.
11548 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11549 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11550 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11551 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11552 (c6x_preferred_simd_mode): Likewise.
11553 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11554 (epiphany_rtx_costs): Likewise.
11555 * config/epiphany/predicates.md (proper_comparison_operator):
11557 * config/frv/frv.c (condexec_memory_operand): Likewise.
11558 (frv_emit_move): Likewise.
11559 (output_move_single): Likewise.
11560 (output_condmove_single): Likewise.
11561 (frv_hard_regno_mode_ok): Likewise.
11562 (frv_matching_accg_mode): Likewise.
11563 * config/h8300/h8300.c (split_adds_subs): Likewise.
11564 (h8300_rtx_costs): Likewise.
11565 (h8300_print_operand): Likewise.
11566 (compute_mov_length): Likewise.
11567 (output_logical_op): Likewise.
11568 (compute_logical_op_length): Likewise.
11569 (compute_logical_op_cc): Likewise.
11570 (h8300_shift_needs_scratch_p): Likewise.
11571 (output_a_shift): Likewise.
11572 (compute_a_shift_length): Likewise.
11573 (compute_a_shift_cc): Likewise.
11574 (expand_a_rotate): Likewise.
11575 (output_a_rotate): Likewise.
11576 * config/i386/i386.c (classify_argument): Likewise.
11577 (function_arg_advance_32): Likewise.
11578 (function_arg_32): Likewise.
11579 (function_arg_64): Likewise.
11580 (function_value_64): Likewise.
11581 (ix86_gimplify_va_arg): Likewise.
11582 (ix86_legitimate_constant_p): Likewise.
11583 (put_condition_code): Likewise.
11584 (split_double_mode): Likewise.
11585 (ix86_avx256_split_vector_move_misalign): Likewise.
11586 (ix86_expand_vector_logical_operator): Likewise.
11587 (ix86_split_idivmod): Likewise.
11588 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11589 (ix86_build_const_vector): Likewise.
11590 (ix86_build_signbit_mask): Likewise.
11591 (ix86_match_ccmode): Likewise.
11592 (ix86_cc_modes_compatible): Likewise.
11593 (ix86_expand_branch): Likewise.
11594 (ix86_expand_sse_cmp): Likewise.
11595 (ix86_expand_sse_movcc): Likewise.
11596 (ix86_expand_int_sse_cmp): Likewise.
11597 (ix86_expand_vec_perm_vpermi2): Likewise.
11598 (ix86_expand_vec_perm): Likewise.
11599 (ix86_expand_sse_unpack): Likewise.
11600 (ix86_expand_int_addcc): Likewise.
11601 (ix86_split_to_parts): Likewise.
11602 (ix86_vectorize_builtin_gather): Likewise.
11603 (ix86_vectorize_builtin_scatter): Likewise.
11604 (avx_vpermilp_parallel): Likewise.
11605 (inline_memory_move_cost): Likewise.
11606 (ix86_tieable_integer_mode_p): Likewise.
11607 (x86_maybe_negate_const_int): Likewise.
11608 (ix86_expand_vector_init_duplicate): Likewise.
11609 (ix86_expand_vector_init_one_nonzero): Likewise.
11610 (ix86_expand_vector_init_one_var): Likewise.
11611 (ix86_expand_vector_init_concat): Likewise.
11612 (ix86_expand_vector_init_interleave): Likewise.
11613 (ix86_expand_vector_init_general): Likewise.
11614 (ix86_expand_vector_set): Likewise.
11615 (ix86_expand_vector_extract): Likewise.
11616 (emit_reduc_half): Likewise.
11617 (ix86_emit_i387_round): Likewise.
11618 (ix86_mangle_type): Likewise.
11619 (ix86_expand_round_sse4): Likewise.
11620 (expand_vec_perm_blend): Likewise.
11621 (canonicalize_vector_int_perm): Likewise.
11622 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11623 (expand_vec_perm_1): Likewise.
11624 (expand_vec_perm_interleave3): Likewise.
11625 (expand_vec_perm_even_odd_pack): Likewise.
11626 (expand_vec_perm_even_odd_1): Likewise.
11627 (expand_vec_perm_broadcast_1): Likewise.
11628 (ix86_vectorize_vec_perm_const_ok): Likewise.
11629 (ix86_expand_vecop_qihi): Likewise.
11630 (ix86_expand_mul_widen_hilo): Likewise.
11631 (ix86_expand_sse2_abs): Likewise.
11632 (ix86_expand_pextr): Likewise.
11633 (ix86_expand_pinsr): Likewise.
11634 (ix86_preferred_simd_mode): Likewise.
11635 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11636 * config/i386/sse.md (*andnot<mode>3): Likewise.
11637 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11638 (*<code><mode>3): Likewise.
11639 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11640 (ia64_expand_atomic_op): Likewise.
11641 (ia64_arg_type): Likewise.
11642 (ia64_mode_to_int): Likewise.
11643 (ia64_scalar_mode_supported_p): Likewise.
11644 (ia64_vector_mode_supported_p): Likewise.
11645 (expand_vec_perm_broadcast): Likewise.
11646 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11647 (iq2000_function_arg_advance): Likewise.
11648 (iq2000_function_arg): Likewise.
11649 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11650 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11651 (m68k_libcall_value): Likewise.
11652 (m68k_function_value): Likewise.
11653 (sched_attr_op_type): Likewise.
11654 * config/mcore/mcore.c (mcore_output_move): Likewise.
11655 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11657 (microblaze_function_arg): Likewise.
11658 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11659 (mips_print_operand): Likewise.
11660 (mips_mode_ok_for_mov_fmt_p): Likewise.
11661 (mips_vector_mode_supported_p): Likewise.
11662 (mips_preferred_simd_mode): Likewise.
11663 (mips_expand_vpc_loongson_even_odd): Likewise.
11664 (mips_expand_vec_unpack): Likewise.
11665 (mips_expand_vi_broadcast): Likewise.
11666 (mips_expand_vector_init): Likewise.
11667 (mips_expand_vec_reduc): Likewise.
11668 (mips_expand_msa_cmp): Likewise.
11669 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11670 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11671 (cc_flags_for_mode): Likewise.
11672 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11673 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11674 (nds32_output_casesi_pc_relative): Likewise.
11675 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11676 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11677 (nvptx_gen_unpack): Likewise.
11678 (nvptx_gen_pack): Likewise.
11679 (nvptx_gen_shuffle): Likewise.
11680 (nvptx_gen_wcast): Likewise.
11681 (nvptx_preferred_simd_mode): Likewise.
11682 * config/pa/pa.c (pa_secondary_reload): Likewise.
11683 * config/pa/predicates.md (base14_operand): Likewise.
11684 * config/powerpcspe/powerpcspe-c.c
11685 (altivec_resolve_overloaded_builtin): Likewise.
11686 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11688 (rs6000_preferred_simd_mode): Likewise.
11689 (output_vec_const_move): Likewise.
11690 (rs6000_expand_vector_extract): Likewise.
11691 (rs6000_split_vec_extract_var): Likewise.
11692 (reg_offset_addressing_ok_p): Likewise.
11693 (rs6000_legitimate_offset_address_p): Likewise.
11694 (rs6000_legitimize_address): Likewise.
11695 (rs6000_emit_set_const): Likewise.
11696 (rs6000_const_vec): Likewise.
11697 (rs6000_emit_move): Likewise.
11698 (spe_build_register_parallel): Likewise.
11699 (rs6000_darwin64_record_arg_recurse): Likewise.
11700 (swap_selector_for_mode): Likewise.
11701 (spe_init_builtins): Likewise.
11702 (paired_init_builtins): Likewise.
11703 (altivec_init_builtins): Likewise.
11704 (do_load_for_compare): Likewise.
11705 (rs6000_generate_compare): Likewise.
11706 (rs6000_expand_float128_convert): Likewise.
11707 (emit_load_locked): Likewise.
11708 (emit_store_conditional): Likewise.
11709 (rs6000_output_function_epilogue): Likewise.
11710 (rs6000_handle_altivec_attribute): Likewise.
11711 (rs6000_function_value): Likewise.
11712 (emit_fusion_gpr_load): Likewise.
11713 (emit_fusion_p9_load): Likewise.
11714 (emit_fusion_p9_store): Likewise.
11715 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11716 (fusion_gpr_mem_load): Likewise.
11717 (fusion_addis_mem_combo_load): Likewise.
11718 (fusion_addis_mem_combo_store): Likewise.
11719 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11720 (fusion_gpr_mem_load): Likewise.
11721 (fusion_addis_mem_combo_load): Likewise.
11722 (fusion_addis_mem_combo_store): Likewise.
11723 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11725 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11726 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11727 (rs6000_preferred_simd_mode): Likewise.
11728 (output_vec_const_move): Likewise.
11729 (rs6000_expand_vector_extract): Likewise.
11730 (rs6000_split_vec_extract_var): Likewise.
11731 (reg_offset_addressing_ok_p): Likewise.
11732 (rs6000_legitimate_offset_address_p): Likewise.
11733 (rs6000_legitimize_address): Likewise.
11734 (rs6000_emit_set_const): Likewise.
11735 (rs6000_const_vec): Likewise.
11736 (rs6000_emit_move): Likewise.
11737 (rs6000_darwin64_record_arg_recurse): Likewise.
11738 (swap_selector_for_mode): Likewise.
11739 (paired_init_builtins): Likewise.
11740 (altivec_init_builtins): Likewise.
11741 (rs6000_expand_float128_convert): Likewise.
11742 (emit_load_locked): Likewise.
11743 (emit_store_conditional): Likewise.
11744 (rs6000_output_function_epilogue): Likewise.
11745 (rs6000_handle_altivec_attribute): Likewise.
11746 (rs6000_function_value): Likewise.
11747 (emit_fusion_gpr_load): Likewise.
11748 (emit_fusion_p9_load): Likewise.
11749 (emit_fusion_p9_store): Likewise.
11750 * config/rx/rx.c (rx_gen_move_template): Likewise.
11751 (flags_from_mode): Likewise.
11752 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11753 (s390_slb_comparison): Likewise.
11754 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11755 (s390_vector_mode_supported_p): Likewise.
11756 (s390_cc_modes_compatible): Likewise.
11757 (s390_match_ccmode_set): Likewise.
11758 (s390_canonicalize_comparison): Likewise.
11759 (s390_emit_compare_and_swap): Likewise.
11760 (s390_branch_condition_mask): Likewise.
11761 (s390_rtx_costs): Likewise.
11762 (s390_secondary_reload): Likewise.
11763 (__SECONDARY_RELOAD_CASE): Likewise.
11764 (s390_expand_cs): Likewise.
11765 (s390_preferred_simd_mode): Likewise.
11766 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11767 * config/sh/sh.c (sh_print_operand): Likewise.
11768 (dump_table): Likewise.
11769 (sh_secondary_reload): Likewise.
11770 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11771 * config/sh/sh.md (casesi_worker_1): Likewise.
11772 (casesi_worker_2): Likewise.
11773 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11774 (fcc_comparison_operator): Likewise.
11775 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11776 (emit_soft_tfmode_cvt): Likewise.
11777 (sparc_preferred_simd_mode): Likewise.
11778 (output_cbranch): Likewise.
11779 (sparc_print_operand): Likewise.
11780 (sparc_expand_vec_perm_bmask): Likewise.
11781 (vector_init_bshuffle): Likewise.
11782 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11783 (spu_vector_mode_supported_p): Likewise.
11784 (spu_expand_insv): Likewise.
11785 (spu_emit_branch_or_set): Likewise.
11786 (spu_handle_vector_attribute): Likewise.
11787 (spu_builtin_splats): Likewise.
11788 (spu_builtin_extract): Likewise.
11789 (spu_builtin_promote): Likewise.
11790 (spu_expand_sign_extend): Likewise.
11791 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11792 (tilegx_simd_int): Likewise.
11793 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11794 (tilepro_simd_int): Likewise.
11795 * config/v850/v850.c (const_double_split): Likewise.
11796 (v850_print_operand): Likewise.
11797 (ep_memory_offset): Likewise.
11798 * config/vax/vax.c (vax_rtx_costs): Likewise.
11799 (vax_output_int_move): Likewise.
11800 (vax_output_int_add): Likewise.
11801 (vax_output_int_subtract): Likewise.
11802 * config/visium/predicates.md (visium_branch_operator): Likewise.
11803 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11804 (visium_print_operand_address): Likewise.
11805 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11806 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11807 (xtensa_expand_conditional_branch): Likewise.
11808 (xtensa_copy_incoming_a7): Likewise.
11809 (xtensa_output_literal): Likewise.
11810 * dfp.c (decimal_real_maxval): Likewise.
11811 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11814 Alan Hayward <alan.hayward@arm.com>
11815 David Sherwood <david.sherwood@arm.com>
11817 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11818 (mode_nunits_inline): Likewise.
11819 (mode_inner_inline): Likewise.
11820 (mode_unit_size_inline): Likewise.
11821 (mode_unit_precision_inline): Likewise.
11822 (emit_insn_modes_h): Likewise. Also emit a #define of the
11824 (emit_mode_wider): Add an E_ prefix to mode names.
11825 (emit_mode_complex): Likewise.
11826 (emit_mode_inner): Likewise.
11827 (emit_mode_adjustments): Likewise.
11828 (emit_mode_int_n): Likewise.
11829 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11830 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11831 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11832 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11833 (CRC32_BUILTIN, ENTRY): Likewise.
11834 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11835 (aarch64_pop_regs): Likewise.
11836 (aarch64_process_components): Likewise.
11837 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11838 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11839 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11840 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11841 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11842 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11844 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11845 (pcmpistr): Likewise.
11846 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11847 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11848 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11850 * config/rl78/rl78.c (mduc_regs): Likewise.
11851 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11852 (htm_expand_builtin): Likewise.
11853 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11854 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11855 * config/xtensa/xtensa.c (print_operand): Likewise.
11856 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11857 (NUM_MODE_VECTOR_INT): Likewise.
11858 * genoutput.c (null_operand): Likewise.
11859 (output_operand_data): Likewise.
11860 * genrecog.c (print_parameter_value): Likewise.
11861 * lra.c (debug_operand_data): Likewise.
11863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11865 * dwarf2out.c (scompare_loc_descriptor_wide)
11866 (scompare_loc_descriptor_narrow): New functions, split out from...
11867 (scompare_loc_descriptor): ...here.
11868 * expmed.c (emit_store_flag_int): New function, split out from...
11869 (emit_store_flag): ...here.
11871 2017-08-30 Richard Biener <rguenther@suse.de>
11873 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11874 (dwarf2out_early_finish): Move setting of AT_pubnames from
11875 early debug output to early finish.
11877 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11879 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11880 and -mdata-region to the assembler.
11882 2017-08-30 Richard Biener <rguenther@suse.de>
11884 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11886 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11887 (dwarf2out_early_global_decl): Only generate a DIE for the
11888 abstract origin if it doesn't already exist or is a declaration DIE.
11889 (resolve_addr): Do not add the linkage name twice when
11890 generating a stub DIE for the DW_TAG_GNU_call_site target.
11892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11894 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11895 Use machine_mode rather than int for arg1_mode.
11897 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11900 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11901 that the second argument of the built-in functions to unpack
11902 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11903 switch statement instead a lot of if statements.
11904 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11905 Allow 64-bit values to be in Altivec registers as well as
11906 traditional floating point registers.
11907 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11909 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11911 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11912 MAX_REGS_PER_ADDRESS == 1.
11914 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11916 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11917 * config/i386/i386.c (ix86_option_override_internal): Simplify
11918 setting of opts->x_flag_entry.
11920 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11921 Jakub Jelinek <jakub@redhat.com>
11922 Richard Biener <rguenther@suse.de>
11924 PR tree-optimization/81503
11925 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11926 folded constant fits in the target type; reorder tests for clarity.
11928 2017-08-29 Martin Liska <mliska@suse.cz>
11930 * passes.def: Include pass_lower_switch.
11931 * stmt.c (dump_case_nodes): Remove and move to
11932 tree-switch-conversion.
11933 (case_values_threshold): Likewise.
11934 (expand_switch_as_decision_tree_p): Likewise.
11935 (emit_case_decision_tree): Likewise.
11936 (expand_case): Likewise.
11937 (balance_case_nodes): Likewise.
11938 (node_has_low_bound): Likewise.
11939 (node_has_high_bound): Likewise.
11940 (node_is_bounded): Likewise.
11941 (emit_case_nodes): Likewise.
11942 (struct simple_case_node): New struct.
11943 (add_case_node): Remove.
11944 (emit_case_dispatch_table): Use vector instead of case_list.
11945 (reset_out_edges_aux): Remove.
11946 (compute_cases_per_edge): Likewise.
11947 (expand_case): Build list of simple_case_node.
11948 (expand_sjlj_dispatch_table): Use it.
11949 * tree-switch-conversion.c (struct case_node): Moved from
11950 stmt.c and adjusted.
11951 (emit_case_nodes): Likewise.
11952 (node_has_low_bound): Likewise.
11953 (node_has_high_bound): Likewise.
11954 (node_is_bounded): Likewise.
11955 (case_values_threshold): Likewise.
11956 (reset_out_edges_aux): Likewise.
11957 (compute_cases_per_edge): Likewise.
11958 (add_case_node): Likewise.
11959 (dump_case_nodes): Likewise.
11960 (balance_case_nodes): Likewise.
11961 (expand_switch_as_decision_tree_p): Likewise.
11962 (emit_jump): Likewise.
11963 (emit_case_decision_tree): Likewise.
11964 (try_switch_expansion): Likewise.
11965 (do_jump_if_equal): Likewise.
11966 (emit_cmp_and_jump_insns): Likewise.
11967 (fix_phi_operands_for_edge): New function.
11968 (record_phi_operand_mapping): Likewise.
11969 (class pass_lower_switch): New pass.
11970 (pass_lower_switch::execute): New function.
11971 (make_pass_lower_switch): Likewise.
11972 (conditional_probability):
11973 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11974 * tree-pass.h: Add make_pass_lower_switch.
11976 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11979 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11982 2017-08-29 Richard Biener <rguenther@suse.de>
11984 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11985 we do not add a DW_AT_inline attribute twice.
11986 (gen_subprogram_die): Remove code setting DW_AT_inline on
11987 DECL_ABSTRACT_P nodes.
11989 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11991 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11992 calls to internal functions.
11993 (gimplify_modify_expr): Likewise.
11994 * tree-call-cdce.c (use_internal_fn): Likewise.
11995 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11996 (convert_to_divmod): Set the nothrow flag.
11997 * tree-if-conv.c (predicate_mem_writes): Likewise.
11998 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11999 (vectorizable_call): Likewise.
12000 (vectorizable_store): Likewise.
12001 (vectorizable_load): Likewise.
12002 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12003 (vect_recog_mask_conversion_pattern): Likewise.
12005 2017-08-29 Martin Liska <mliska@suse.cz>
12008 * gcc.c (driver_handle_option): Add new argument.
12009 * opts-common.c (handle_option): Pass
12010 target_option_override_hook.
12011 * opts-global.c (lang_handle_option): Add new option.
12012 (set_default_handlers): Add new argument.
12013 (decode_options): Likewise.
12014 * opts.c (target_handle_option): Likewise.
12015 (common_handle_option): Call target_option_override_hook.
12016 * opts.h (struct cl_option_handler_func): Add hook for
12017 target option override.
12018 (struct cl_option_handlers): Likewise.
12019 (set_default_handlers): Add new argument.
12020 (decode_options): Likewise.
12021 (common_handle_option): Likewise.
12022 (target_handle_option): Likewise.
12023 * toplev.c (toplev::main): Pass targetm.target_option.override
12026 2017-08-29 Richard Biener <rguenther@suse.de>
12027 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12029 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12030 life to the active subtree.
12032 2017-08-28 Jeff Law <law@redhat.com>
12034 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12035 derive_equivalences.
12036 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12037 Code moved into....
12038 (edge_info::derive_equivalences): New private member function
12040 * tree-ssa-dom.c (class edge_info): Changed from a struct
12041 to a class. Add ctor/dtor, methods and data members.
12042 (edge_info::edge_info): Renamed from allocate_edge_info.
12043 Initialize additional members.
12044 (edge_info::~edge_info): New.
12045 (free_dom_edge_info): Delete the edge info.
12046 (record_edge_info): Use new class & associated member functions.
12047 Tighten forms for testing for edge equivalences.
12048 (record_temporary_equivalences): Iterate over the simple
12049 equivalences rather than assuming there's only one per edge.
12050 (cprop_into_successor_phis): Iterate over the simple
12051 equivalences rather than assuming there's only one per edge.
12052 (optimize_stmt): Use operand_equal_p rather than pointer
12053 equality for mini-DSE code.
12055 2017-08-28 Nathan Sidwell <nathan@acm.org>
12057 * gcc.c (execute): Fold SIGPIPE handling into switch
12058 statement. Adjust internal error message.
12060 2017-08-28 Richard Biener <rguenther@suse.de>
12063 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12064 Do nothing for removed DIEs.
12066 2017-08-28 Richard Biener <rguenther@suse.de>
12068 PR tree-optimization/81977
12069 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12072 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12075 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12076 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12077 using targetm.gen_mem_thread_fence.
12079 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12082 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12083 predicate to register_operand. Reorder operands.
12084 (*btr<mode>): Ditto.
12085 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12086 (*btr<mode>_mask): Ditto.
12088 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12090 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12091 * config/rs6000/xmmintrin.h: New file.
12092 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12094 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12097 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12098 parameter and_insn and return it.
12099 (recombine_lvx_pattern): Insert a copy to ensure availability of
12100 the base register of the copied masking operation at the point of
12101 the instruction replacement.
12102 (recombine_stvx_pattern): Likewise.
12104 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12106 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12107 undocumented switches.
12108 (-mpower9-dform-vector): Likewise.
12109 (-mpower9-dform): Likewise.
12110 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12111 comments to delete references to -mpower9-dform* switches.
12112 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12113 Delete reference to -mpower9-dform* switches, test for
12114 -mpower9-vector instead.
12115 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12116 (OTHER_P9_VECTOR_MASKS): Likewise.
12117 (POWERPC_MASKS): Likewise.
12118 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12119 tests against -mpower9-dform* to -mpower9-vector. Delete code
12120 that checked for -mpower9-dform* consistancy with other options.
12121 Add test for -mpower9-misc to enable other power9 switches.
12122 (rs6000_init_hard_regno_mode_ok): Likewise.
12123 (rs6000_option_override_internal): Likewise.
12124 (rs6000_emit_prologue): Likewise.
12125 (rs6000_emit_epilogue): Likewise.
12126 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12127 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12128 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12130 (emit_fusion_p9_store): Likewise.
12131 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12132 resetting these macros if the assembler does not support ISA 3.0
12134 (TARGET_P9_DFORM_VECTOR): Likewise.
12135 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12136 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12138 2017-08-25 Alan Modra <amodra@gmail.com>
12140 PR rtl-optimization/81747
12141 * cse.c (cse_extended_basic_block): Don't attempt to record
12142 equivalences for degenerate conditional branches that branch
12143 to their fall-through.
12145 2017-08-24 Martin Sebor <msebor@redhat.com>
12147 PR middle-end/81908
12148 * gimple-fold.c (size_must_be_zero_p): New function.
12149 (gimple_fold_builtin_memory_op): Call it.
12151 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12153 * config/rs6000/mm_malloc.h: New file.
12155 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12157 PR tree-optimization/81913
12158 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12159 analysis when either IVs in condition can wrap.
12161 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12163 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12164 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12166 2017-08-24 Richard Biener <rguenther@suse.de>
12169 * targhooks.c (default_target_can_inline_p): Properly
12170 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12171 is present and always compare.
12172 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12173 infer -mfpmath=sse from TARGET_SSE_P.
12174 (ix86_can_inline_p): Properly use target_option_default_node when
12175 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12177 2017-08-24 Richard Biener <rguenther@suse.de>
12180 * dwarf2out.c (output_die): Handle flag_generate_offload like
12182 (output_comp_unit): Likewise.
12183 (gen_array_type_die): Likewise.
12184 (dwarf2out_early_finish): Likewise.
12185 (note_variable_value_in_expr): Likewise.
12186 (dwarf2out_finish): Likewise. Adjust assert.
12187 * cgraphunit.c (symbol_table::compile): Move setting of
12188 flag_generate_offload earlier ...
12189 (symbol_table::finalize_compilation_unit): ... here, before
12190 early debug finalization.
12192 2017-08-24 Richard Biener <rguenther@suse.de>
12194 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12195 and ipa-fnsummary.h.
12196 (ix86_can_inline_p): When ix86_fpmath flags do not match
12197 check whether the callee uses FP math at all.
12199 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12201 PR middle-end/81931
12202 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12203 instead of TYPE_PRECISION.
12205 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12206 Alan Hayward <alan.hayward@arm.com>
12207 David Sherwood <david.sherwood@arm.com>
12209 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12211 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12213 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12215 (ix86_frame::stack_realign_allocate): New field.
12216 (struct machine_frame_state): Modify comments.
12217 (machine_frame_state::sp_realigned_fp_end): New field.
12218 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12219 layout calculation.
12220 (sp_valid_at): Add assertion to assure no attempt to access invalid
12221 offset of a realigned stack.
12222 (fp_valid_at): Likewise.
12223 (choose_baseaddr): Modify comments.
12224 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12225 ix86_expand_prologue.
12226 (ix86_expand_prologue): Modify stack realignment and allocation.
12227 (ix86_expand_epilogue): Modify comments.
12228 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12229 avx2_runtime, avx512f, and avx512f_runtime.
12231 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12233 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12234 (mstackrealign): Do not init to -1.
12235 * config/i386/i386.c (ix86_option_override_internal):
12236 Check opts_set, not opts when setting default value of
12237 opts->x_ix86_force_align_arg_pointer.
12239 2017-08-23 Richard Biener <rguenther@suse.de>
12241 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12242 lang_hooks.decl_printable_name.
12243 * print-rtl-function.c (print_rtx_function): Likewise.
12244 * tree-pretty-print.c (dump_function_header): Likewise.
12246 2017-08-23 Richard Biener <rguenther@suse.de>
12249 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12250 -g0 at compile-time.
12252 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12254 PR middle-end/19706
12255 * doc/sourcebuild.texi (Other hardware attributes):
12258 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12260 PR middle-end/19706
12261 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12262 Fix single-use check.
12264 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12266 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12268 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12270 * doc/install.texi: Modify to add more details on running selected
12273 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12275 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12276 is combined with -mabi=ms.
12277 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12280 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12282 PR tree-optimization/81488
12283 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12284 and cached_basis fields.
12285 (MAX_SPREAD): New constant.
12286 (alloc_cand_and_find_basis): Initialize new fields.
12287 (clear_visited): New function.
12288 (create_phi_basis_1): Rename from create_phi_basis, set visited
12289 and cached_basis fields.
12290 (create_phi_basis): New wrapper function.
12291 (phi_add_costs_1): Rename from phi_add_costs, add spread
12292 parameter, set visited field, short-circuit when limits reached.
12293 (phi_add_costs): New wrapper function.
12294 (record_phi_increments_1): Rename from record_phi_increments, set
12296 (record_phi_increments): New wrapper function.
12297 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12298 (phi_incr_cost): New wrapper function.
12299 (all_phi_incrs_profitable_1): Rename from
12300 all_phi_incrs_profitable, set visited field.
12301 (all_phi_incrs_profitable): New wrapper function.
12303 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12304 Alan Hayward <alan.hayward@arm.com>
12305 David Sherwood <david.sherwood@arm.com>
12307 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12308 that takes the outer and inner modes.
12309 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12310 comparison as the canonical test for a paradoxical subreg.
12311 * combine.c (simplify_set): Use paradoxical_subreg_p.
12312 (make_extraction): Likewise.
12313 (force_to_mode): Likewise.
12314 (rtx_equal_for_field_assignment_p): Likewise.
12315 (gen_lowpart_for_combine): Likewise.
12316 (simplify_comparison): Likewise.
12317 * cse.c (equiv_constant): Likewise.
12318 * expmed.c (store_bit_field_1): Likewise.
12319 * final.c (alter_subreg): Likewise.
12320 * fwprop.c (propagate_rtx): Likewise.
12321 (forward_propagate_subreg): Likewise.
12322 * ira-conflicts.c (ira_build_conflicts): Likewise.
12323 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12324 * lra-constraints.c (curr_insn_transform): Likewise.
12325 (split_reg): Likewise.
12326 * lra-eliminations.c (move_plus_up): Likewise.
12327 (lra_eliminate_regs_1): Likewise.
12328 * recog.c (general_operand): Likewise.
12329 * ree.c (combine_reaching_defs): Likewise.
12330 * reload.c (push_reload): Likewise.
12331 (find_reloads): Likewise.
12332 * reload1.c (elimination_effects): Likewise.
12333 (compute_reload_subreg_offset): Likewise.
12334 (choose_reload_regs): Likewise.
12335 * rtlanal.c (subreg_lsb_1): Likewise.
12336 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12337 (simplify_subreg): Likewise.
12338 * var-tracking.c (track_loc_p): Likewise.
12339 * emit-rtl.c (byte_lowpart_offset): Likewise.
12340 (paradoxical_subreg_p): Delete out-of-line definition.
12342 2017-08-22 Jeff Law <law@redhat.com>
12344 PR tree-optimization/81741
12345 PR tree-optimization/71947
12346 * tree-ssa-dom.c: Include tree-inline.h.
12347 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12348 equivalences if one is more expensive to compute than the other.
12349 * tree-ssa-scopedtables.h (class const_or_copies): Make
12350 record_const_or_copy_raw method private.
12351 (class avail_exprs_stack): New method simplify_binary_operation.
12352 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12353 avail_exprs_stack::simplify_binary_operation as needed.
12354 (avail_exprs_stack::simplify_binary_operation): New function.
12356 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12358 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12359 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12360 (DOT_SYMBOLS): Likewise.
12361 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12362 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12363 (RS6000_ABI_NAME): Likewise.
12364 (TARGET_CMODEL): Likewise.
12365 (TOC_SECTION_ASM_OP): Likewise.
12366 (SET_CMODEL): New macro.
12367 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12369 2017-08-22 Richard Biener <rguenther@suse.de>
12371 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12372 to allow for free-lang-data replacements similar to verify_type_variant.
12374 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12375 Alan Hayward <alan.hayward@arm.com>
12376 David Sherwood <david.sherwood@arm.com>
12378 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12379 VOIDmode for the LABEL_REF.
12381 2017-08-22 Richard Biener <rguenther@suse.de>
12383 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12385 2017-08-22 Richard Biener <rguenther@suse.de>
12387 * common.opt (feliminate-dwarf2-dups): Ignore.
12388 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12389 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12390 same_die_p_wrap, compute_section_prefix,
12391 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12392 (comdat_symbol_id, comdat_symbol_number): Likewise.
12393 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12395 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12396 (output_die): Mark unreachable path unreachable.
12397 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12398 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12399 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12400 (dwarf2out_early_finish): Likewise.
12402 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12404 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12406 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12409 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12410 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12411 (avr_attribute_table) <io, io_low, address>: Initialize
12412 .decl_required with true.
12414 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12416 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12417 undocumented debugging options.
12418 (-mvsx-scalar-double): Likewise.
12419 (-mallow-df-permute): Likewise.
12420 (-mvectorize-builtins): Likewise.
12421 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12422 (rs6000_builtin_vectorized_function): Likewise.
12423 (rs6000_builtin_md_vectorized_function): Likewise.
12424 (rs6000_opt_vars): Likewise.
12426 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12429 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12430 (*btrq_imm): Rename from *btrq.
12431 (*btcq_imm): Rename from *btcq.
12432 (btsc): New code attribute.
12433 (*<btsc><mode>): New insn pattern.
12434 (*btr<mode>): Ditto.
12435 (*<btsc><mode>_mask): New insn_and_split pattern.
12436 (*btr<mode>_mask): Ditto.
12438 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12439 Alan Hayward <alan.hayward@arm.com>
12440 David Sherwood <david.sherwood@arm.com>
12442 * function.c (pad_below): Simplify padding calculation.
12444 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12445 Alan Hayward <alan.hayward@arm.com>
12446 David Sherwood <david.sherwood@arm.com>
12448 * target.def (function_prologue): Remove frame size argument.
12449 (function_epilogue): Likewise.
12450 * doc/tm.texi: Regenerate.
12451 * final.c (final_start_function): Update call to function_prologue.
12452 (final_end_function): Update call to function_epilogue.
12453 (default_function_pro_epilogue): Remove frame size argument.
12454 * output.h (default_function_pro_epilogue): Likewise.
12455 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12456 (arm_output_function_prologue): Likewise.
12457 * config/frv/frv.c (frv_function_prologue): Likewise.
12458 (frv_function_epilogue): Likewise.
12459 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12460 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12461 (ia64_output_function_epilogue): Likewise.
12462 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12463 (m32r_output_function_epilogue): Likewise.
12464 * config/microblaze/microblaze.c (microblaze_function_prologue)
12465 (microblaze_function_epilogue): Likewise.
12466 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12467 (mips_output_function_epilogue): Likewise.
12468 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12469 (mmix_target_asm_function_epilogue): Likewise.
12470 * config/msp430/msp430.c (msp430_start_function): Likewise.
12471 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12472 (nds32_asm_function_epilogue): Likewise.
12473 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12474 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12475 (pa_output_function_epilogue): Likewise.
12476 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12477 (rs6000_output_function_epilogue): Likewise.
12478 * config/rl78/rl78.c (rl78_start_function): Likewise.
12479 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12480 (rs6000_output_function_epilogue): Likewise.
12481 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12482 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12483 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12484 (sparc_asm_function_epilogue): Likewise.
12486 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12488 * tree.h (type_has_mode_precision_p): New function.
12489 * convert.c (convert_to_integer_1): Use it.
12490 * expr.c (expand_expr_real_2): Likewise.
12491 (expand_expr_real_1): Likewise.
12492 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12493 * match.pd: Likewise.
12494 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12495 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12496 * tree-tailcall.c (process_assignment): Likewise.
12497 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12498 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12499 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12500 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12501 (vectorizable_assignment): Likewise.
12502 (vectorizable_shift): Likewise.
12503 (vectorizable_operation): Likewise.
12504 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12506 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12508 * match.pd: Add pow (C, x) simplification.
12510 2017-08-21 Richard Biener <rguenther@suse.de>
12512 PR tree-optimization/81900
12513 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12514 for blocks with abnormal predecessors.
12515 (compute_antic): Do not set visited flag prematurely.
12517 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12520 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12522 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12524 * stor-layout.h (vector_type_mode): Move to...
12525 * tree.h (vector_type_mode): ...here.
12526 * stor-layout.c (vector_type_mode): Move to...
12527 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12529 2017-08-21 Richard Biener <rguenther@suse.de>
12531 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12532 register_external_die hooks.
12533 (debug_false_tree_charstarstar_uhwistar): Declare.
12534 (debug_nothing_tree_charstar_uhwi): Likewise.
12535 * debug.c (do_nothing_debug_hooks): Adjust.
12536 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12537 (debug_nothing_tree_charstar_uhwi): Likewise.
12538 * dbxout.c (dbx_debug_hooks): Adjust.
12539 (xcoff_debug_hooks): Likewise.
12540 * sdbout.c (sdb_debug_hooks): Likewise.
12541 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12542 * dwarf2out.c (macinfo_label_base): New global.
12543 (dwarf2out_register_external_die): New function for the
12544 register_external_die hook.
12545 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12546 (dwarf2_debug_hooks): Use them.
12547 (dwarf2_lineno_debug_hooks): Adjust.
12548 (struct die_struct): Add with_offset flag.
12549 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12550 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12551 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12552 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12553 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12554 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12555 defining section names for the early LTO debug variants.
12556 (reset_indirect_string): New helper.
12557 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12558 (print_dw_val): Add support for offsetted symbol references.
12559 (get_ultimate_context): Split out from is_cxx.
12560 (is_cxx): Use get_ultimate_context.
12561 (is_fortran): Add decl overload.
12562 (compute_comp_unit_symbol): Split out worker from
12563 compute_section_prefix.
12564 (compute_section_prefix): Call compute_comp_unit_symbol and
12565 set comdat_type_p here.
12566 (output_die): Skip DIE symbol output for the LTO added one.
12567 Handle DIE symbol references with offset.
12568 (output_comp_unit): Guard section name mangling properly.
12569 For LTO debug sections emit a symbol at the section beginning
12570 which we use to refer to its DIEs.
12571 (add_abstract_origin_attribute): For DIEs registered via
12572 dwarf2out_register_external_die directly refer to the early
12573 DIE rather than indirectly through the shadow one we created.
12574 Remove obsolete call to dwarf2out_abstract_function for
12575 non-function/block origins.
12576 (gen_array_type_die): When generating early LTO debug do
12577 not emit DW_AT_string_length.
12578 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12579 late when in LTO. As suggested place a gcc_unreachable for
12580 the DECL_ABSTRACT_P case.
12581 (gen_subprogram_die): Avoid another specification DIE
12582 for early built declarations/definitions for the late LTO case.
12583 (gen_variable_die): Add type references for late duplicated VLA dies
12585 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12586 we have the abstract instance already.
12587 (process_scope_var): Adjust decl DIE contexts in LTO which
12588 first puts them in limbo.
12589 (gen_decl_die): Do not generate type DIEs late apart from
12590 types for VLAs or for decls we do not yet have a DIE. Do not
12591 call dwarf2out_abstract_function late.
12592 (dwarf2out_early_global_decl): Make sure to create DIEs
12593 for abstract instances of a decl first.
12594 (dwarf2out_late_global_decl): Adjust comment.
12595 (output_macinfo_op): With multiple macro sections use
12596 macinfo_label_base to distinguish labels.
12597 (output_macinfo): Likewise. Update macinfo_label_base.
12598 Pass in the line info label.
12599 (note_variable_value_in_expr): When generating LTO resolve
12600 all variable values here by generating DIEs as needed.
12601 (init_sections_and_labels): Add early LTO debug flag parameter
12602 and generate different sections and names if set. Add generation
12603 counter for the labels so we can have multiple of them.
12604 (reset_dies): Helper to allow DIEs to be output multiple times.
12605 (dwarf2out_finish): When outputting DIEs to the fat part of an
12606 LTO object first reset DIEs.
12607 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12608 (modified_type_die): Check for decl_ultimate_origin being self
12610 (gen_type_die_with_usage): Likewise.
12611 (gen_typedef_die): Allow decl_ultimate_origin being self.
12612 (set_decl_abstract_flags): Remove.
12613 (set_block_abstract_flags): Likewise.
12614 (dwarf2out_abstract_function): Treat the early generated DIEs
12615 as the abstract copy and only add DW_AT_inline and
12616 DW_AT_artificial here and call set_decl_origin_self.
12617 If the DIE has an abstract origin don't do anything.
12618 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12619 if we have none yet (Go fails to build one, PR78628).
12620 (variably_modified_type_p): Prevent endless recursion for Ada
12621 cyclic pointer types.
12622 * lto-streamer-in.c: Include debug.h.
12623 (dref_queue): New global.
12624 (lto_read_tree_1): Stream in DIE references.
12625 (lto_input_tree): Register DIE references.
12626 (input_function): Stream DECL_DEBUG_ARGS.
12627 * lto-streamer-out.c: Include debug.h.
12628 (lto_write_tree_1): Output DIE references.
12629 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12630 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12631 (output_function): Stream DECL_DEBUG_ARGS.
12632 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12633 Stream DECL_ABSTRACT_ORIGIN.
12634 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12635 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12636 DECL_CONTEXT for file-scope decls.
12637 * lto-streamer.h (struct dref_entry): Declare.
12638 (dref_queue): Likewise.
12639 * cfgexpand.c (pass_expand::execute): Do not call the
12640 outlining_inline_function hook here.
12641 * lto-wrapper.c (debug_obj): New global.
12642 (tool_cleanup): Unlink it if required.
12643 (debug_objcopy): New function.
12644 (run_gcc): Handle early debug sections in the IL files by
12645 extracting them to separate files, partially linkin them and
12646 feeding the result back as result to the linker.
12647 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12648 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12649 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12650 sections into a separate segment.
12651 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12653 (darwin_asm_dwarf_section): Likewise.
12654 (darwin_asm_output_dwarf_offset): Likewise.
12655 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12657 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12658 Alan Hayward <alan.hayward@arm.com>
12659 David Sherwood <david.sherwood@arm.com>
12661 * read-md.h (md_reader::record_potential_iterator_use): Replace
12662 pointer argument with an rtx and an index.
12663 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12664 (apply_mode_iterator): Likewise.
12665 (apply_code_iterator): Likewise.
12666 (apply_int_iterator): Likewise.
12667 (apply_subst_iterator): Likewise.
12668 (record_iterator_use): Likewise.
12669 (record_attribute_use): Likewise.
12670 (md_reader::record_potential_iterator_use): Likewise. Update calls
12671 to record_iterator_use and apply_iterator.
12672 (iterator_use): Replace ptr with x and index.
12673 (attribute_use): Likewise.
12674 (apply_attribute_uses): Update calls to apply_iterator.
12675 (apply_iterators): Likewise. Update initialization of iterator_use.
12676 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12677 and record_potential_iterator_use.
12678 (rtx_reader::read_rtx_operand): Likewise.
12680 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12681 Alan Hayward <alan.hayward@arm.com>
12682 David Sherwood <david.sherwood@arm.com>
12684 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12687 2017-08-21 Richard Biener <rguenther@suse.de>
12689 PR middle-end/81884
12690 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12691 at struct end conservatively when comparing common bases.
12693 2017-08-21 Richard Biener <rguenther@suse.de>
12695 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12696 (mem_ref_in_stmt): Remove.
12697 (determine_max_movement): Use ref index to get at the reference.
12698 (invariantness_dom_walker::before_dom_children): Deal with
12699 lim data already initialized.
12700 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12702 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12704 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12705 (TARGET_ISA_ROUND): Ditto.
12706 (TARGET_ROUND): Ditto.
12707 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12708 * config/i386/i386.md: Ditto.
12709 * config/i386/sse.md: Ditto.
12710 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12711 with OPTION_MASK_ISA_SSE4_1.
12713 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12716 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12717 __builtin_ia32_lzcnt_u16.
12719 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12722 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12723 (rs6000_set_current_function): Rewrite function to use it.
12725 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12728 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12729 and TYPE_WARN_IF_NOT_ALIGN.
12730 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12731 (handle_warn_if_not_align): New.
12732 (place_union_field): Call handle_warn_if_not_align.
12733 (place_field): Call handle_warn_if_not_align.
12734 Copy TYPE_WARN_IF_NOT_ALIGN.
12735 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12736 (layout_type): Likewise.
12737 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12739 (tree_decl_common): Add warn_if_not_align.
12740 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12741 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12742 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12743 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12744 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12745 * doc/extend.texi: Document warn_if_not_aligned attribute.
12746 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12748 2017-08-17 Martin Liska <mliska@suse.cz>
12751 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12752 (get_data_dependence): Use it as pointer type.
12753 (distribute_loop): Likewise.
12755 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12757 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12758 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12759 (p8_vmrgow_<mode>_direct): New define_insn.
12760 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12761 handle endianness for vmrgew and vmrgow permute patterns.
12763 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12765 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12766 * config/rs6000/rs6000-cpus.def: Remove comment.
12767 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12768 (POWERPC_MASKS): Likewise.
12769 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12770 use of TARGET_VSX_TIMODE.
12771 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12772 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12773 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12774 (rs6000_option_override_internal): Remove dead code.
12775 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12776 (rs6000_legitimize_reload_address): Likewise.
12777 (rs6000_legitimate_address_p): Likewise.
12778 (rs6000_opt_masks): Delete "vsx-timode".
12779 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12780 from function comment.
12781 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12782 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12783 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12785 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12786 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12787 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12788 (VSX_TI): Likewise.
12790 (define_peephole2): Likewise.
12792 2017-08-17 Martin Sebor <msebor@redhat.com>
12795 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12796 past the end of an array.
12797 (test_pp_format): Add test cases.
12799 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12801 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12802 missing ECF_NOTHROW flags.
12804 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12807 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12808 operands residing in integer registers.
12809 (*vsx_le_perm_load_<mode>): Likewise.
12810 (*vsx_le_perm_store_<mode>): Likewise.
12811 (define_peephole2): Add peepholes to optimize the above.
12813 2017-08-17 Marek Polacek <polacek@redhat.com>
12815 PR middle-end/81814
12816 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12817 to mimic what shorten_compare did. Change the return type to bool.
12818 (fold_cond_expr_with_comparison): Update call to
12819 operand_equal_for_comparison_p.
12820 (fold_ternary_loc): Likewise.
12822 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12824 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12826 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12828 2017-08-17 Richard Biener <rguenther@suse.de>
12830 * tree-ssa-structalias.c (solve_graph): When propagating
12831 to successors update the graphs succ edges and avoid duplicate work.
12833 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12836 * config/i386/i386.c (ix86_option_override_internal): Save target
12837 specific options after ix86_stack_protector_guard_reg was changed.
12839 2017-08-17 Richard Biener <rguenther@suse.de>
12841 PR tree-optimization/81827
12842 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12843 (new_var_info): Initialize it conservatively.
12844 (get_call_vi): Mark register vars.
12845 (new_scalar_tmp_constraint_exp): Likewise.
12846 (handle_rhs_call): Likewise.
12847 (handle_const_call): Likewise.
12848 (create_function_info_for): Likewise.
12849 (solve_constraints): Sort varinfos to separate register from
12850 non-register vars to pack points-to solution bitmaps during
12853 2017-08-17 Marek Polacek <polacek@redhat.com>
12855 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12857 2017-08-17 Richard Biener <rguenther@suse.de>
12859 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12860 to true when overflow is undefined and we saturated the result.
12862 2017-08-17 Alan Modra <amodra@gmail.com>
12865 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12866 Don't use store multiple if only one reg needs saving.
12867 (interesting_frame_related_regno): New function.
12868 (rs6000_frame_related): Don't emit frame info for regs that
12870 (rs6000_emit_epilogue): Likewise.
12872 2017-08-16 Nathan Sidwell <nathan@acm.org>
12874 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12875 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12876 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12877 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12878 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12879 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12880 (verify_type): Adjust for TYPE_BINFO move.
12881 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12882 process TYPE_BINFO directly.
12883 (hash_tree): Likewise.
12884 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12886 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12889 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12891 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12893 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12896 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12897 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12898 Add ix86_binary_operator_ok to insn constraint.
12899 (*iordi_1_bts): Ditto.
12900 (*xordi_1_btc): Ditto.
12901 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12902 Update corresponding peephole2 pattern.
12906 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12908 PR tree-optimization/81832
12909 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12910 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12912 2017-08-16 Marek Polacek <polacek@redhat.com>
12915 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12916 regarding size_zero_node.
12918 2017-08-16 Martin Liska <mliska@suse.cz>
12921 * config.gcc: Respect previously set extra_objs in case
12924 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12926 PR tree-optimization/81835
12927 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12928 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12929 not depend on the phi.
12931 2017-08-16 Alan Modra <amodra@gmail.com>
12933 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12936 2017-08-16 Alan Modra <amodra@gmail.com>
12938 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12939 (save_reg_p): ..into this. Update all callers.
12940 (first_reg_to_save): Simplify.
12942 2017-08-16 Alan Modra <amodra@gmail.com>
12944 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12947 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12951 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12952 constant count if that count is less than 32.
12954 2017-08-15 Nathan Sidwell <nathan@acm.org>
12956 * gcc.c (execute): Emit friendlier message if inferior is killed
12957 by an external cause.
12959 2017-08-15 Richard Biener <rguenther@suse.de>
12961 PR tree-optimization/81790
12962 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12963 CONSTRUCTORs from simplifying and VN.
12965 2017-08-14 Martin Sebor <msebor@redhat.com>
12967 * builtin-attrs.def: Add comments.
12969 2017-08-14 Martin Sebor <msebor@redhat.com>
12972 * doc/extend.texi (attribute nonstring): Document new attribute.
12974 2017-08-14 Martin Sebor <msebor@redhat.com>
12977 * tree-diagnostic.c (default_tree_printer): Handle %G.
12978 * gimple-pretty-print.h (percent_G_format): Declare new function.
12979 * gimple-pretty-print.c (percent_G_format): Define.
12980 * tree-pretty-print.c (percent_K_format): Add argument.
12982 2017-08-14 Martin Sebor <msebor@redhat.com>
12984 PR translation/79998
12985 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12986 Remove a stray space.
12988 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12991 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12992 (*iordi_1_bts): Ditto.
12993 (*xordi_1_btc): Ditto.
12995 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12998 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12999 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13001 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13002 quoted strings, and make more translator-friendly.
13003 (darwin_rs6000_override_options): Likewise.
13004 (rs6000_option_override_internal): Likewise.
13005 (rs6000_return_in_memory): Fix overlong line.
13006 (init_cmulative_args): Use quoted strings, and make more
13007 translator-friendly.
13008 (rs6000_pass_by_reference): Fix overlong line.
13009 (def_builtin): Use quoted strings.
13010 (altivec_expand_predicate_builtin): Use quoted strings, and make
13011 more translator-friendly.
13012 (htm_expand_builtin): Use quoted strings.
13013 (cpu_expand_builtin): Use quoted strings, and make more
13014 translator-friendly.
13015 (altivec_expand_builtin): Likewise.
13016 (paired_expand_predicate_builtin): Likewise.
13017 (rs6000_invalid_builtin): Likewise.
13018 (builtin_function_type): Use quoted strings.
13019 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13020 more translator-friendly.
13021 (rs6000_trampoline_init): Likewise.
13022 (rs6000_handle_altivec_attribute): Likewise.
13023 (rs6000_inner_target_options): Use quoted strings.
13024 (rs6000_disable_incompatible_switches): Likewise.
13025 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13026 strings, and make more translator-friendly.
13027 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13029 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13031 PR tree-optimization/81799
13032 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13033 cond_expr to simple gimple operand.
13035 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13037 PR middle-end/46932
13038 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13040 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13044 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13045 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13046 TARGET_GASISR_PROLOGUES.
13047 * config/avr/avr.c (avr_option_override): Same.
13048 (avr_pass_pre_proep::execute): Same.
13050 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13053 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13054 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13056 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13058 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13059 define_insn_and_split. Split to a memory load from 0 in
13060 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13061 using PTR mode iterator.
13062 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13063 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13064 (*add_tp_<mode>): Redefine as define_insn_and_split.
13065 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13066 address space. Merge with *add_tp_x32 using PTR mode iterator.
13067 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13068 Split to an add with a memory load from 0 in
13069 DEFAULT_TLS_SEG_REG address space.
13071 2017-08-12 Andrew Pinski <apinski@cavium.com>
13073 * config/aarch64/aarch64-option-extensions.def (rdma):
13074 Fix feature string to what Linux prints out in /proc/cpuinfo.
13076 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13079 * dwarf2out.c (modified_type_die): For C typedef types that have
13080 an ultimate origin, process the ultimate origin instead of the
13082 (gen_typedef_die): Assert that input DECLs have no ultimate
13084 (gen_type_die_with_usage): For typedef variants that have an
13085 ultimate origin, just call gen_decl_die on the original DECL.
13086 (process_scope_var): Avoid creating DIEs for local typedefs and
13087 concrete static variables.
13089 2017-08-12 Alan Modra <amodra@gmail.com>
13093 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13094 match gnu-user.h startfile.
13095 (ENDFILE_LINUX_SPEC): Similarly.
13097 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13100 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13102 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13104 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13105 * config/aarch64/aarch64.md (mov<mode>): Change.
13106 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13107 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13108 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13110 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13112 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13113 for storage order barriers.
13115 2017-08-11 Martin Liska <mliska@suse.cz>
13118 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13119 variables of void type.
13121 2017-08-11 Martin Liska <mliska@suse.cz>
13123 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13124 TARGET_SUPPORTS_ALIASES.
13125 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13126 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13127 (optimize_weakref): Likewise.
13128 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13129 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13130 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13132 2017-08-11 Martin Liska <mliska@suse.cz>
13135 * config/i386/i386.c (make_resolver_func): Do complete
13136 refactoring of the function.
13138 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13141 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13142 * config/i386/i386.c (ix86_stack_protect_guard): Use
13143 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13144 * doc/invoke.texi (x86 Options): Document
13145 -mstack-protector-guard-symbol= option.
13147 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13149 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13150 * config/i386/i386.c (ix86_split_stack_guard): New function.
13151 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13152 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13153 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13154 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13155 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13156 (split_stack_space_check): Call ix86_split_stack_guard.
13158 2017-08-10 Martin Sebor <msebor@redhat.com>
13160 * print-tree.c (print_node): Print location using the established
13162 Replace spaces with colons.
13163 (debug_raw, debug): Ditto.
13165 2017-08-10 Martin Sebor <msebor@redhat.com>
13168 * pretty-print.c (pp_format): Correct the handling of %s precision.
13170 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13173 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13175 (ix86_finalize_stack_frame_flags): This. Also clear
13176 frame_pointer_needed if -fno-omit-frame-pointer is used without
13178 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13179 with ix86_finalize_stack_frame_flags.
13180 (ix86_expand_epilogue): Likewise.
13181 (ix86_expand_split_stack_prologue): Likewise.
13182 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13184 2017-08-10 Martin Liska <mliska@suse.cz>
13187 * c-attribs.c (handle_target_attribute):
13188 Report warning for an empty string argument of target attribute.
13190 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13193 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13195 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13196 or DECL_NONLOCAL labels.
13197 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13198 or DECL_NONLOCAL labels here.
13200 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13202 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13203 to indicate when early gimple folding has been disabled.
13204 (rs6000_gimple_fold_builtin): Add debug content.
13205 (rs6000_invalid_builtin): Fix whitespace.
13206 (rs6000_expand_builtin): Fix whitespace.
13207 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13209 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13212 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13213 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13215 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13217 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13218 (qdf24xx): Likewise.
13219 * config/aarch64/aarch64.md: Include falkor.md.
13220 * config/aarch64/falkor.md: New.
13222 2017-08-09 Marek Polacek <polacek@redhat.com>
13225 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13226 * diagnostic.c (emit_diagnostic): Add a comment.
13227 (emit_diagnostic_valist): New function.
13229 2017-08-09 Marek Polacek <polacek@redhat.com>
13232 * input.c (make_location): New overload.
13233 * input.h (make_location): Declare.
13235 2017-08-08 Alan Modra <amodra@gmail.com>
13236 H.J. Lu <hongjiu.lu@intel.com>
13239 * gcc.c (NO_PIE_SPEC): Delete.
13240 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13242 (LINK_PIE_SPEC): ..to here.
13243 (LINK_COMMAND_SPEC): Support -no-pie.
13244 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13245 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13247 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13248 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13249 (ENDFILE_CRTEND_SPEC): Similarly.
13251 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13254 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13255 (mstack-protector-guard-offset=): Ditto.
13256 * config/i386/i386.c (ix86_option_override): Handle
13257 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13259 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13260 ix86_stack_protect_guard_offset variables.
13261 (TARGET_STACK_PROTECT_GUARD): Always define.
13262 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13263 and -mstack-protector-guard-offset= options.
13265 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13267 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13268 boundary case for the last candidate.
13270 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13272 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13273 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13275 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13277 PR middle-end/19706
13278 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13279 * config/aarch64/aarch64-builtins.c
13280 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13281 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13282 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13284 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13285 Andrew Pinski <pinskia@gmail.com>
13287 PR middle-end/19706
13288 * internal-fn.def (XORSIGN): New.
13289 * optabs.def (xorsign_optab): New.
13290 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13291 (convert_expand_mult_copysign): New.
13292 (pass_optimize_widening_mul::execute): Call
13293 convert_expand_mult_copysign.
13295 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13297 PR tree-optimization/81354
13298 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13299 Insert on edges rather than explicitly creating landing pads.
13300 (analyze_candidates_and_replace): Commit edge inserts.
13302 2017-08-08 Richard Biener <rguenther@suse.de>
13304 PR middle-end/81719
13305 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13306 (expand_simple_operations): Also look through ADDR_EXPRs with
13307 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13309 2017-08-08 Richard Biener <rguenther@suse.de>
13311 PR tree-optimization/81723
13312 * tree-vect-slp.c (struct bst_traits): New hash traits.
13313 (bst_fail): New global.
13314 (vect_build_slp_tree_2): New worker, split out from ...
13315 (vect_build_slp_tree): ... this now wrapping it with using
13316 bst_fail set to cache SLP tree build fails. Properly handle
13318 (vect_analyze_slp_instance): Allocate and free bst_fail.
13320 2017-08-08 Martin Liska <mliska@suse.cz>
13323 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13324 LABEL_DECLs that can be from a different function.
13326 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13328 PR tree-optimization/81744
13329 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13330 loop's number of iterations.
13332 2017-08-08 Martin Liska <mliska@suse.cz>
13334 * asan.c: Include header files.
13335 * attribs.c (build_decl_attribute_variant): New function moved
13337 (build_type_attribute_qual_variant): Likewise.
13338 (cmp_attrib_identifiers): Likewise.
13339 (simple_cst_list_equal): Likewise.
13340 (omp_declare_simd_clauses_equal): Likewise.
13341 (attribute_value_equal): Likewise.
13342 (comp_type_attributes): Likewise.
13343 (build_type_attribute_variant): Likewise.
13344 (lookup_ident_attribute): Likewise.
13345 (remove_attribute): Likewise.
13346 (merge_attributes): Likewise.
13347 (merge_type_attributes): Likewise.
13348 (merge_decl_attributes): Likewise.
13349 (merge_dllimport_decl_attributes): Likewise.
13350 (handle_dll_attribute): Likewise.
13351 (attribute_list_equal): Likewise.
13352 (attribute_list_contained): Likewise.
13353 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13354 (lookup_attribute_by_prefix): Likewise.
13355 * bb-reorder.c: Include header files.
13356 * builtins.c: Likewise.
13357 * calls.c: Likewise.
13358 * cfgexpand.c: Likewise.
13359 * cgraph.c: Likewise.
13360 * cgraphunit.c: Likewise.
13361 * convert.c: Likewise.
13362 * dwarf2out.c: Likewise.
13363 * final.c: Likewise.
13364 * fold-const.c: Likewise.
13365 * function.c: Likewise.
13366 * gimple-expr.c: Likewise.
13367 * gimple-fold.c: Likewise.
13368 * gimple-pretty-print.c: Likewise.
13369 * gimple.c: Likewise.
13370 * gimplify.c: Likewise.
13371 * hsa-common.c: Likewise.
13372 * hsa-gen.c: Likewise.
13373 * internal-fn.c: Likewise.
13374 * ipa-chkp.c: Likewise.
13375 * ipa-cp.c: Likewise.
13376 * ipa-devirt.c: Likewise.
13377 * ipa-fnsummary.c: Likewise.
13378 * ipa-inline.c: Likewise.
13379 * ipa-visibility.c: Likewise.
13381 * lto-cgraph.c: Likewise.
13382 * omp-expand.c: Likewise.
13383 * omp-general.c: Likewise.
13384 * omp-low.c: Likewise.
13385 * omp-offload.c: Likewise.
13386 * omp-simd-clone.c: Likewise.
13387 * opts-global.c: Likewise.
13388 * passes.c: Likewise.
13389 * predict.c: Likewise.
13390 * sancov.c: Likewise.
13391 * sanopt.c: Likewise.
13392 * symtab.c: Likewise.
13393 * toplev.c: Likewise.
13394 * trans-mem.c: Likewise.
13395 * tree-chkp.c: Likewise.
13396 * tree-eh.c: Likewise.
13397 * tree-into-ssa.c: Likewise.
13398 * tree-object-size.c: Likewise.
13399 * tree-parloops.c: Likewise.
13400 * tree-profile.c: Likewise.
13401 * tree-ssa-ccp.c: Likewise.
13402 * tree-ssa-live.c: Likewise.
13403 * tree-ssa-loop.c: Likewise.
13404 * tree-ssa-sccvn.c: Likewise.
13405 * tree-ssa-structalias.c: Likewise.
13406 * tree-ssa.c: Likewise.
13407 * tree-streamer-in.c: Likewise.
13408 * tree-vectorizer.c: Likewise.
13409 * tree-vrp.c: Likewise.
13410 * tsan.c: Likewise.
13411 * ubsan.c: Likewise.
13412 * varasm.c: Likewise.
13413 * varpool.c: Likewise.
13414 * tree.c: Remove functions moved to attribs.[ch].
13415 * tree.h: Likewise.
13416 * config/aarch64/aarch64.c: Add attrs.h header file.
13417 * config/alpha/alpha.c: Likewise.
13418 * config/arc/arc.c: Likewise.
13419 * config/arm/arm.c: Likewise.
13420 * config/avr/avr.c: Likewise.
13421 * config/bfin/bfin.c: Likewise.
13422 * config/c6x/c6x.c: Likewise.
13423 * config/cr16/cr16.c: Likewise.
13424 * config/cris/cris.c: Likewise.
13425 * config/darwin.c: Likewise.
13426 * config/epiphany/epiphany.c: Likewise.
13427 * config/fr30/fr30.c: Likewise.
13428 * config/frv/frv.c: Likewise.
13429 * config/ft32/ft32.c: Likewise.
13430 * config/h8300/h8300.c: Likewise.
13431 * config/i386/winnt.c: Likewise.
13432 * config/ia64/ia64.c: Likewise.
13433 * config/iq2000/iq2000.c: Likewise.
13434 * config/lm32/lm32.c: Likewise.
13435 * config/m32c/m32c.c: Likewise.
13436 * config/m32r/m32r.c: Likewise.
13437 * config/m68k/m68k.c: Likewise.
13438 * config/mcore/mcore.c: Likewise.
13439 * config/microblaze/microblaze.c: Likewise.
13440 * config/mips/mips.c: Likewise.
13441 * config/mmix/mmix.c: Likewise.
13442 * config/mn10300/mn10300.c: Likewise.
13443 * config/moxie/moxie.c: Likewise.
13444 * config/msp430/msp430.c: Likewise.
13445 * config/nds32/nds32-isr.c: Likewise.
13446 * config/nds32/nds32.c: Likewise.
13447 * config/nios2/nios2.c: Likewise.
13448 * config/nvptx/nvptx.c: Likewise.
13449 * config/pa/pa.c: Likewise.
13450 * config/pdp11/pdp11.c: Likewise.
13451 * config/powerpcspe/powerpcspe.c: Likewise.
13452 * config/riscv/riscv.c: Likewise.
13453 * config/rl78/rl78.c: Likewise.
13454 * config/rx/rx.c: Likewise.
13455 * config/s390/s390.c: Likewise.
13456 * config/sh/sh.c: Likewise.
13457 * config/sol2.c: Likewise.
13458 * config/sparc/sparc.c: Likewise.
13459 * config/spu/spu.c: Likewise.
13460 * config/stormy16/stormy16.c: Likewise.
13461 * config/tilegx/tilegx.c: Likewise.
13462 * config/tilepro/tilepro.c: Likewise.
13463 * config/v850/v850.c: Likewise.
13464 * config/vax/vax.c: Likewise.
13465 * config/visium/visium.c: Likewise.
13466 * config/xtensa/xtensa.c: Likewise.
13468 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13471 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13472 constraints since the -mupper-regs-* switches have been
13474 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13475 into a vector from a double word element that was extracted from
13476 another vector, and eliminate extra XXPERMDI instructions.
13477 (vsx_concat_<mode>_2): Likewise.
13478 (vsx_concat_<mode>_3): Likewise.
13479 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13480 concat to allow optimizing inserts from previous extracts.
13482 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13484 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13485 memory reference to a SSP offset in TLS address space.
13486 (ix86_print_operand) <case '@'>: Remove.
13487 (ix86_print_operand_punct_valid_p): Remove '@' code.
13488 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13489 UNSPEC_SP_TLS_TEST.
13490 (stack_tls_protect_set_<mode>): Remove.
13491 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13492 (stack_tls_protect_test_<mode>): Remove.
13493 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13495 2017-08-07 Olivier Hainque <hainque@adacore.com>
13498 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13500 2017-08-07 Douglas Rupp <rupp@adacore.com>
13502 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13503 variable was referenced as multidir in command.
13505 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13508 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13511 2017-08-07 Martin Liska <mliska@suse.cz>
13513 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13514 * config/rl78/rl78.c: Add include of attribs.h.
13515 * config/sh/sh.c: Likewise.
13516 * config/v850/v850.c: Likewise.
13518 2017-08-07 Tom de Vries <tom@codesourcery.com>
13520 PR middle-end/78266
13521 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13523 2017-08-07 Martin Liska <mliska@suse.cz>
13525 * config/mips/mips.c: Include attribs.h.
13527 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13530 * doc/invoke.texi: Document change in behvaior for -Ofast for
13533 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13535 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13537 (aarch64_pop_regs): Likewise.
13538 (aarch64_gen_load_pair): Likewise.
13539 (aarch64_save_callee_saves): Likewise.
13540 (aarch64_restore_callee_saves): Likewise.
13542 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13544 * config/i386/i386.c: Revert the last change.
13546 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13549 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13551 (ix86_finalize_stack_frame_flags): This. Also clear
13552 frame_pointer_needed if -fno-omit-frame-pointer is used without
13554 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13555 with ix86_finalize_stack_frame_flags.
13556 (ix86_expand_epilogue): Likewise.
13557 (ix86_expand_split_stack_prologue): Likewise.
13559 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13562 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13563 to P_AES for Westmere.
13565 2017-08-07 Jonathan Yong <10walls@gmail.com>
13567 * config/i386/mingw.opt (fset-stack-executable): Removed.
13568 * config/i386/cygming.opt (fset-stack-executable): Moved
13570 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13572 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13574 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13576 2017-08-07 Marek Polacek <polacek@redhat.com>
13578 PR middle-end/81737
13579 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13581 2017-08-07 Martin Liska <mliska@suse.cz>
13583 * attribs.h (canonicalize_attr_name): New function.
13584 (cmp_attribs): Move from c-format.c and adjusted.
13585 (is_attribute_p): Moved from tree.h.
13586 * tree-inline.c: Add new includes.
13587 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13588 (private_is_attribute_p): Remove.
13589 (private_lookup_attribute): Likewise.
13590 (private_lookup_attribute_by_prefix): Simplify.
13591 (remove_attribute): Use is_attribute_p.
13592 * tree.h: Remove removed declarations.
13594 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13596 PR middle-end/81698
13597 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13598 instead of computing it in the function. Formatting fix.
13599 (expand_case): Don't rely on default_edge being the first edge,
13600 clear it if removing it, pass default_edge to
13601 emit_case_dispatch_table.
13602 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13605 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13607 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13608 insn in the function, emit NOP after the insn.
13610 2017-08-06 Tom de Vries <tom@codesourcery.com>
13612 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13615 2017-08-06 Tom de Vries <tom@codesourcery.com>
13617 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13620 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13622 PR tree-optimization/57371
13623 * match.pd: New pattern.
13625 2017-08-04 Marek Polacek <polacek@redhat.com>
13627 PR middle-end/81695
13628 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13629 perform the computation in offset_int.
13631 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13633 PR tree-optimization/81136
13634 * tree-vectorizer.h: Include tree-hash-traits.h.
13635 (vec_base_alignments): New typedef.
13636 (vec_info): Add a base_alignments field.
13637 (vect_record_base_alignments): Declare.
13638 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13640 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13641 (create_data_ref): Add an is_conditional_in_stmt argument.
13642 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13643 the is_conditional_in_stmt field.
13644 (data_ref_loc): Add an is_conditional_in_stmt field.
13645 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13646 (find_data_references_in_stmt): Update call to create_data_ref.
13647 (graphite_find_data_references_in_stmt): Likewise.
13648 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13649 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13650 (vect_record_base_alignment): New function.
13651 (vect_record_base_alignments): Likewise.
13652 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13653 for nested statements even if we fail to compute a misalignment.
13654 Use pooled base alignments for unconditional references.
13655 (vect_find_same_alignment_drs): Compare base addresses instead
13657 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13658 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13660 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13662 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13663 Add an explicit name for the enum. Use auto_vec for slp_instances
13664 and grouped_stores.
13665 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13667 (_bb_vec_info): Add a constructor and destructor.
13668 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13669 (destroy_loop_vec_info): Delete.
13670 (vect_destroy_datarefs): Likewise.
13671 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13672 (vec_info::vec_info): New function.
13673 (vec_info::~vec_info): Likewise.
13674 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13675 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13676 destroy_loop_vec_info.
13677 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13678 (_loop_vec_info::_loop_vec_info): ...this.
13679 (destroy_loop_vec_info): Replace with...
13680 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13681 the stmt_vec_infos. Leave handling of vec_info information to its
13682 destructor. Remove explicit vector releases.
13683 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13684 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13685 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13686 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13687 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13688 (destroy_bb_vec_info): Replace with...
13689 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13690 information to its destructor.
13691 (vect_slp_analyze_bb_1): Use new and delete instead of
13692 new_bb_vec_info and destroy_bb_vec_info.
13693 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13696 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13698 * tree-data-ref.h (subscript): Add access_fn field.
13699 (data_dependence_relation): Add could_be_independent_p.
13700 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13701 (same_access_functions): Move to tree-data-ref.c.
13702 * tree-data-ref.c (ref_contains_union_access_p): New function.
13703 (access_fn_component_p): Likewise.
13704 (access_fn_components_comparable_p): Likewise.
13705 (dr_analyze_indices): Add a reference to access_fn_component_p.
13706 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13708 (constant_access_functions): Likewise.
13709 (add_other_self_distances): Likewise.
13710 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13711 (initialize_data_dependence_relation): Use XCNEW and remove
13712 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13713 of access functions that have the same type. Allow the
13714 subsequence to end with different bases in some circumstances.
13715 Record the chosen access functions in SUB_ACCESS_FN.
13716 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13717 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13718 (subscript_dependence_tester_1): Likewise dra and drb.
13719 (build_classic_dist_vector): Update calls accordingly.
13720 (subscript_dependence_tester): Likewise.
13721 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13722 DDR_COULD_BE_INDEPENDENT_P.
13723 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13724 comp_alias_ddrs instead of may_alias_ddrs.
13725 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13727 (vect_analyze_data_ref_dependence): Use it if
13728 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13729 distance vectors if that fails.
13730 (dependence_distance_ge_vf): New function.
13731 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13732 LOOP_VINFO_MAY_ALIAS_DDRS.
13734 2017-08-04 Richard Biener <rguenther@suse.de>
13736 PR middle-end/81705
13737 * fold-const.c (fold_binary_loc): Properly restrict
13738 minus_var0 && minus_var1 case when associating undefined overflow
13741 2017-08-04 Tom de Vries <tom@codesourcery.com>
13743 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13745 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13747 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13748 Don't start diagnostic messages with a capital letter.
13749 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13751 (rs6000_invalid_builtin): Likewise.
13752 (rs6000_trampoline_init): Likewise.
13754 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13757 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13758 after setting changeable df flags.
13760 2017-08-03 Richard Biener <rguenther@suse.de>
13762 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13763 up if the use is in USE - X.
13765 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13767 * toplev.c (dumpfile.h): New include.
13768 (internal_error_reentered): New static function. Use it...
13769 (internal_error_function): ...here to handle reentered internal_error.
13771 2017-08-03 Richard Biener <rguenther@suse.de>
13773 PR middle-end/81148
13774 * fold-const.c (split_tree): Add minus_var and minus_con
13775 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13776 here but always use minus_*.
13777 (associate_trees): Assert we never associate with MINUS_EXPR
13778 and NULL first operand. Do not recurse for PLUS_EXPR operands
13779 when associating as MINUS_EXPR either.
13780 (fold_binary_loc): Track minus_var and minus_con.
13782 2017-08-03 Tom de Vries <tom@codesourcery.com>
13785 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13786 ACCEL_COMPILER, apply finish_options on
13787 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13789 2017-08-03 Tom de Vries <tom@codesourcery.com>
13792 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13793 function_entry_patch_area_size > 0.
13795 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13798 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13799 instead of 10??LU, perform unit multiplication in wide_int,
13800 don't change alloc_object_size_limit if the limit is larger
13803 PR tree-optimization/81655
13804 PR tree-optimization/81588
13805 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13806 the case when ranges[i].low and high are 1 for unsigned type with
13809 PR middle-end/81052
13810 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13811 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13813 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13815 * tree-vrp.h: Add include guard.
13817 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13820 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13821 (ud2): New insn pattern.
13822 * config/i386/i386.c (ix86_expand_epilogue):
13823 For naked functions, generate ud2 instead of trap insn.
13825 2017-08-02 Marek Polacek <polacek@redhat.com>
13828 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13830 2017-08-02 Tom de Vries <tom@codesourcery.com>
13831 Cesar Philippidis <cesar@codesourcery.com>
13833 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13834 Add missing edge probabilities.
13836 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13838 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13839 Correct endianness.
13841 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13843 PR middle-end/79499
13844 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13845 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13846 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13847 prologue_seq sequences - if any.
13849 2017-08-02 Richard Biener <rguenther@suse.de>
13851 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13852 via vectors if supported, integer extracts via punning if supported
13853 or otherwise vector extracts.
13855 2017-08-02 Richard Biener <rguenther@suse.de>
13857 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13859 (bitmap_insert_into_set): ... this.
13861 2017-08-02 Richard Biener <rguenther@suse.de>
13863 PR tree-optimization/81633
13865 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13867 PR tree-optimization/71752
13868 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13870 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13872 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13873 (machine_function::call_ms2sysv_pad_out): Remove field.
13874 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13875 (ix86_compute_frame_layout): Likewise.
13877 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13880 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13881 attribute with interrupt attribute.
13883 2017-08-01 Andrew Pinski <apinski@cavium.com>
13885 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13886 BIT_INSERT_EXPR's operand 1
13887 to see if the types precision matches.
13889 2017-08-01 Martin Liska <mliska@suse.cz>
13891 PR middle-end/70140
13892 * builtins.c (expand_builtin_memcpy_args): Remove.
13893 (expand_builtin_memcpy): Call newly added function
13894 expand_builtin_memory_copy_args.
13895 (expand_builtin_memcpy_with_bounds): Likewise.
13896 (expand_builtin_mempcpy): Remove last argument.
13897 (expand_builtin_mempcpy_with_bounds): Likewise.
13898 (expand_builtin_memory_copy_args): New function created from
13899 expand_builtin_mempcpy_args with small modifications.
13900 (expand_builtin_mempcpy_args): Remove.
13901 (expand_builtin_stpcpy): Remove unused argument.
13902 (expand_builtin): Likewise.
13903 (expand_builtin_with_bounds): Likewise.
13905 2017-08-01 Martin Liska <mliska@suse.cz>
13908 Make mempcpy more optimal (PR middle-end/70140).
13910 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13914 __builtin_vec_cmpne verify both arguments are compatible vectors
13915 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13916 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13917 move computation of aligned to after checking the argument types.
13921 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13922 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13925 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13926 Doug Rupp <rupp@adacore.com>
13927 Olivier Hainque <hainque@adacore.com>
13929 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13930 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13932 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13933 for TARGET_OS_CPP_BUILTIN.
13934 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13935 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13937 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13938 passing required abi options to the assembler for EABI configurations.
13939 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13940 of .text.hot and .text.unlikely sections for kernel modules when
13941 using ARM style exceptions.
13942 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13943 options. Add EXTRA_CC1_SPEC.
13944 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13946 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13948 (ARM_TARGET2_DWARF_FORMAT): Define.
13949 * config/arm/t-vxworks: Adjust multilib control to removal of the
13950 Diab command line options.
13952 2017-08-01 Martin Liska <mliska@suse.cz>
13954 PR gcov-profile/81561
13955 * gcov.c (unblock): Make unblocking safe as we need to preserve
13956 index correspondence of blocks and block_lists.
13958 2017-08-01 Richard Biener <rguenther@suse.de>
13960 PR tree-optimization/81181
13961 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13962 (compute_antic): ... end of iteration here.
13964 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13966 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13967 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13968 (ftree-slp-vectorize): Likewise.
13969 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13970 can no longer be set independent of flag_tree_loop_vectorize.
13971 * omp-general.c (emp_max_vf): Likewise.
13972 * opts.c (enable_fdo_optimizations): Remove references to
13973 flag_tree_vectorize, these are now implicit.
13974 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13975 and leave it for the options machinery.
13977 2017-08-01 Martin Liska <mliska@suse.cz>
13979 PR middle-end/70140
13980 * builtins.c (expand_builtin_memcpy_args): Remove.
13981 (expand_builtin_memcpy): Call newly added function
13982 expand_builtin_memory_copy_args.
13983 (expand_builtin_memcpy_with_bounds): Likewise.
13984 (expand_builtin_mempcpy): Remove last argument.
13985 (expand_builtin_mempcpy_with_bounds): Likewise.
13986 (expand_builtin_memory_copy_args): New function created from
13987 expand_builtin_mempcpy_args with small modifications.
13988 (expand_builtin_mempcpy_args): Remove.
13989 (expand_builtin_stpcpy): Remove unused argument.
13990 (expand_builtin): Likewise.
13991 (expand_builtin_with_bounds): Likewise.
13993 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13996 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13997 print "ds:" only for immediates in generic address space.
13999 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14002 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14003 (ix86_function_ok_for_sibcall): Return false for naked functions.
14005 2017-08-01 Richard Biener <rguenther@suse.de>
14007 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14008 (compute_antic): Seed worklist with exit block predecessors.
14009 * cfganal.c (dfs_find_deadend): For a cycle return the source
14010 of the edge closing it.
14012 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14014 * config/aarch64/aarch64.c
14015 (aarch64_can_const_movi_rtx_p): Move 0 check.
14017 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14019 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14020 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14022 * match.pd: Ditto in address comparison pattern.
14024 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14026 PR tree-optimization/81627
14027 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14028 closed ssa form for store-store chain.
14030 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14032 PR tree-optimization/81620
14033 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14034 for store-store chain.
14036 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14038 PR tree-optimization/81588
14039 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14040 ranges[i].in_p, invert comparison code ccode. For >/>=,
14041 swap rhs1 and rhs2 and comparison code unconditionally,
14042 for </<= don't do that. Don't swap rhs1/rhs2 again if
14043 ranges[i].in_p, instead invert comparison code ccode if
14044 opcode or oe->rank is BIT_IOR_EXPR.
14047 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14048 a direct optab to conversion optab.
14049 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14050 with GET_MODE_INNER as last argument instead of optab_handler.
14051 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14052 vector extraction if possible and optab is available.
14053 * expr.c (store_constructor): Use convert_optab_handler instead
14054 of optab_handler. Use vector initialization from smaller
14055 vectors if possible and optab is available.
14056 * tree-vect-stmts.c (vectorizable_load): Likewise.
14057 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14058 now have two modes.
14059 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14060 of vec_init from half-sized vectors with the same element mode.
14061 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14062 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14063 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14064 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14065 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14066 after mode in gen_vec_extract* calls.
14067 (vec_extract<mode>): Renamed to ...
14068 (vec_extract<mode><ssescalarmodelower>): ... this.
14069 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14070 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14071 element mode after mode in gen_vec_init* calls.
14072 (VEC_INIT_HALF_MODE): New mode iterator.
14073 (vec_init<mode>): Renamed to ...
14074 (vec_init<mode><ssescalarmodelower>): ... this.
14075 (vec_init<mode><ssehalfvecmodelower>): New expander.
14076 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14077 (vec_extractv2sfsf): ... this.
14078 (vec_initv2sf): Renamed to ...
14079 (vec_initv2sfsf): ... this.
14080 (vec_extractv2si): Renamed to ...
14081 (vec_extractv2sisi): ... this.
14082 (vec_initv2si): Renamed to ...
14083 (vec_initv2sisi): ... this.
14084 (vec_extractv4hi): Renamed to ...
14085 (vec_extractv4hihi): ... this.
14086 (vec_initv4hi): Renamed to ...
14087 (vec_initv4hihi): ... this.
14088 (vec_extractv8qi): Renamed to ...
14089 (vec_extractv8qiqi): ... this.
14090 (vec_initv8qi): Renamed to ...
14091 (vec_initv8qiqi): ... this.
14092 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14093 (vec_init<mode>): Renamed to ...
14094 (vec_init<mode><VEC_base_l>): ... this.
14095 (vec_extract<mode>): Renamed to ...
14096 (vec_extract<mode><VEC_base_l>): ... this.
14097 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14098 (vec_initv2sfsf): ... this.
14099 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14100 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14101 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14102 element mode after mode in gen_vec_init* calls.
14103 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14104 (vec_init<mode><Vel>): ... this.
14105 (vec_extract<mode>): Renamed to ...
14106 (vec_extract<mode><Vel>): ... this.
14107 * config/aarch64/iterators.md (Vel): New mode attribute.
14108 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14109 Add element mode after mode in gen_vec_extract* calls.
14110 * config/s390/vector.md (non_vec_l): New mode attribute.
14111 (vec_extract<mode>): Renamed to ...
14112 (vec_extract<mode><non_vec_l>): ... this.
14113 (vec_init<mode>): Renamed to ...
14114 (vec_init<mode><non_vec_l>): ... this.
14115 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14116 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14118 * config/arm/iterators.md (V_elem_l): New mode attribute.
14119 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14120 (vec_extract<mode><V_elem_l>): ... this.
14121 (vec_extractv2di): Renamed to ...
14122 (vec_extractv2didi): ... this.
14123 (vec_init<mode>): Renamed to ...
14124 (vec_init<mode><V_elem_l>): ... this.
14125 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14126 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14127 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14128 Add element mode after gen_vec_extract* calls.
14129 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14130 (vec_init<mode><unitmode>): ... this.
14131 (vec_extract<mode>): Renamed to ...
14132 (vec_extract<mode><unitmode>): ... this.
14133 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14134 (vec_init<mode><unitmode>): ... this.
14135 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14136 (vec_initv2sfsf): ... this.
14137 (vec_extractv2sf): Renamed to ...
14138 (vec_extractv2sfsf): ... this.
14139 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14140 Add element mode after gen_vec_extract* calls.
14141 * config/mips/mips.md (unitmode): New mode iterator.
14142 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14143 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14144 * config/spu/spu.md (inner_l): New mode attribute.
14145 (vec_init<mode>): Renamed to ...
14146 (vec_init<mode><inner_l>): ... this.
14147 (vec_extract<mode>): Renamed to ...
14148 (vec_extract<mode><inner_l>): ... this.
14149 * config/sparc/sparc.md (veltmode): New mode iterator.
14150 (vec_init<VMALL:mode>): Renamed to ...
14151 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14152 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14153 (vec_initv2sisi): ... this.
14154 (vec_initv2sf): Renamed to ...
14155 (vec_initv2sfsf): ... this.
14156 (vec_extractv2sf): Renamed to ...
14157 (vec_extractv2sfsf): ... this.
14158 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14159 (vec_init<mode>): Renamed to ...
14160 (vec_init<mode><VEC_base_l>): ... this.
14161 (vec_extract<mode>): Renamed to ...
14162 (vec_extract<mode><VEC_base_l>): ... this.
14163 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14164 (vec_initv2sfsf): ... this.
14165 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14166 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14167 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14168 gen_vec_init* calls.
14170 2017-08-01 Richard Biener <rguenther@suse.de>
14172 PR tree-optimization/81297
14173 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14174 TREE_OVERFLOW from INTEGER_CSTs.
14176 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14178 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14180 2017-07-31 Carl Love <cel@us.ibm.com>
14182 * config/rs6000/rs6000-c: Add support for built-in functions
14183 vector signed char vec_xl_be (signed long long, signed char *);
14184 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14185 vector signed int vec_xl_be (signed long long, signed int *);
14186 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14187 vector signed long long vec_xl_be (signed long long, signed long long *);
14188 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14189 vector signed short vec_xl_be (signed long long, signed short *);
14190 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14191 vector double vec_xl_be (signed long long, double *);
14192 vector float vec_xl_be (signed long long, float *);
14193 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14194 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14195 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14197 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14198 (altivec_expand_builtin): Add switch statement to call
14199 altivec_expand_xl_be for each builtin.
14200 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14201 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14202 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14203 __builtin_vsx_le_be_v16qi.
14204 * doc/extend.texi: Update the built-in documentation file for the
14205 new built-in functions.
14207 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14210 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14212 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14214 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14216 * config.gcc: Add z14.
14217 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14218 CPU model numbers for z13s and z14.
14219 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14221 * config/s390/s390-opts.h (enum processor_type): Rename
14222 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14223 * config/s390/s390.c (processor_table): Add field for CPU name to
14224 be passed to Binutils.
14225 (s390_asm_output_machine_for_arch): Use the new field in
14226 processor_table for Binutils.
14227 (s390_expand_builtin): Replace arch12 with z14.
14228 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14229 (s390_get_sched_attrmask): Likewise.
14230 (s390_get_unit_mask): Likewise.
14231 * config/s390/s390.opt: Add z14 to processor_type enum.
14233 2017-07-31 Martin Jambor <mjambor@suse.cz>
14236 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14237 regardless of optimization level.
14239 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14240 Martin Liska <mliska@suse.cz>
14242 * predict.def: Remove old comment and adjust probability.
14243 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14244 PREDICT statements.
14246 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14249 * config/i386/i386.c (ix86_function_naked): New function.
14250 (ix86_can_use_return_insn_p): Return false for naked functions.
14251 (ix86_expand_prologue): Skip prologue for naked functions.
14252 (ix86_expand_epilogue): Skip epilogue for naked functions
14253 and emit trap instruction.
14254 (ix86_warn_func_return): New function.
14255 (ix86_attribute_table): Add "naked" attribute specification.
14256 (TARGET_WARN_FUNC_RETURN): Define.
14257 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14259 2017-07-31 Martin Liska <mliska@suse.cz>
14261 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14262 (dump_gimple_bb_header): Always dump BB info.
14263 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14265 2017-07-31 Martin Liska <mliska@suse.cz>
14268 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14269 also with current_function_decl non-null equality.
14271 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14274 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14275 change type to the element type, instead add eltype variable and
14276 use it where we are interested in the element type.
14278 PR tree-optimization/81603
14279 * ipa-polymorphic-call.c
14280 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14281 offset arithmetic in offset_int, bail out if the resulting bit offset
14282 doesn't fit into shwi.
14284 2017-07-31 Martin Liska <mliska@suse.cz>
14286 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14287 (gimplify_save_expr): Fix comment.
14289 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14292 * config/i386/i386.c (ix86_function_arg): Update arguments for
14294 (ix86_compute_frame_layout): Set the initial stack offset to
14295 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14296 INCOMING_FRAME_SP_OFFSET.
14297 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14298 stack before exception handler returns.
14299 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14300 the 'ERROR_CODE' for exception handler.
14302 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14304 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14305 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14306 (ASM_OUTPUT_REG_POP): Ditto.
14307 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14308 instead of asm_fprintf to output pure string.
14310 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14312 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14313 to imported_module_or_decl hook.
14314 (debug_nothing_tree_tree_tree_bool): Remove.
14315 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14316 * debug.c (do_nothing_debug_hooks): Use
14317 debug_nothing_tree_tree_tree_bool_bool instead of
14318 debug_nothing_tree_tree_tree_bool.
14319 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14320 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14321 * sdbout.c (sdb_debug_hooks): Likewise.
14322 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14323 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14325 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14326 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14327 attribute, don't add anything.
14329 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14331 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14332 (fold_build2_stat_loc): Likewise.
14333 (fold_build3_stat_loc): Likewise.
14334 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14335 (fold_build1_loc): Remove macro.
14336 (fold_build2_loc): Likewise.
14337 (fold_build3_loc): Likewise.
14339 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14341 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14342 (gimple_build_debug_bind_source_stat): Likewise.
14343 * gimple.h (gimple_build_debug_bind): Remove macro.
14344 (gimple_build_debug_bind_source): Likewise.
14346 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14348 * bitmap.c (bitmap_alloc): Adjust.
14349 (bitmap_gc_alloc): Likewise.
14350 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14352 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14354 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14355 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14356 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14357 (bitmap_gc_alloc_stat): Likewise.
14358 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14362 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14363 * rtl.h (shallow_copy_rtx): Remove macro.
14365 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14367 * emit-rtl.c (gen_raw_REG): Adjust.
14368 * gengenrtl.c (gendef): Likewise.
14369 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14370 * rtl.h (rtx_alloc): Remove macro.
14372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14374 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14375 (build_tree_list_stat): Likewise.
14376 * tree.h (build_tree_list): Remove macro.
14377 (build_tree_list_vec): Likewise.
14379 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14381 * tree.c (make_vector_stat): Remove _stat from name.
14382 (build_vector_stat): Likewise.
14383 * tree.h (make_vector_stat): Remove macro.
14384 (build_vector_stat): Likewise.
14386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14388 * tree.h (build_var_debug_value): Remove prototype.
14390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14392 * tree.c (tree_cons_stat): Remove _stat from name.
14393 * tree.h (tree_cons): Remove macro.
14395 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14397 * tree.c (build_vl_exp_stat): Remove _stat from name.
14398 * tree.h (build_vl_exp): Remove macro.
14400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14402 * tree.c (build_decl_stat): Remove _stat from name.
14403 * tree.h (build_decl): Remove macro.
14405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14407 * gimple.c (gimple_build_with_ops_stat): Adjust.
14408 (gimple_alloc_stat): Remove _stat from name.
14409 * gimple.h (gimple_alloc): Remove macro.
14411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14413 * tree.c (make_tree_vec_stat): Remove _stat from name.
14414 (grow_tree_vec_stat): Likewise.
14415 * tree.h (make_tree_vec_stat): Adjust prototype.
14416 (grow_tree_vec_stat): Likewise.
14417 (make_tree_vec): Remove macro.
14418 (grow_tree_vec): Likewise.
14420 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14422 * fold-const.c (fold_build1_stat_loc): Adjust.
14423 (fold_build2_stat_loc): Likewise.
14424 (fold_build3_stat_loc): Likewise.
14425 * tree.c (build0_stat): Remove _stat from name.
14426 (build1_stat): Likewise.
14427 (build2_stat): Likewise.
14428 (build3_stat): Likewise.
14429 (build4_stat): Likewise.
14430 (build5_stat): Likewise.
14431 * tree.h (build1_loc): Remove macro, and rename _stat function
14433 (build2_loc): Likewise.
14434 (build3_loc): Likewise.
14435 (build4_loc): Likewise.
14436 (build5_loc): Likewise.
14438 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14440 * tree.c (make_int_cst_stat): Remove _stat from name.
14441 * tree.h (make_int_cst_stat): Adjust prototype.
14442 (make_int_cst): Remove macro.
14444 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14446 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14447 * tree.h (make_tree_binfo_stat): Adjust prototype.
14448 (make_tree_binfo): Remove.
14450 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14452 * tree.c (copy_node_stat): Rename to copy_node.
14453 (build_distinct_type_copy): Adjust.
14454 * tree.h (copy_node_stat): Adjust prototype.
14455 (copy_node): Remove macro.
14457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14459 * tree.c (make_node_stat): rename to make_node.
14460 (build_tree_list_stat): Adjust.
14461 (build0_stat): Likewise.
14462 (build2_stat): Likewise.
14463 (build3_stat): Likewise.
14464 (build4_stat): Likewise.
14465 (build5_stat): Likewise.
14466 (build_decl_stat): Likewise.
14467 * tree.h (make_node_stat): Adjust prototype.
14468 (make_node): remove macro.
14470 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14472 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14473 (PPC_FEATURE2_SCV): Likewise.
14474 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14476 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14478 * config/aarch64/aarch64.c
14479 (aarch64_internal_mov_immediate): Add new special pattern.
14480 * config/aarch64/aarch64.md (*movdi_aarch64):
14481 Add reg/32bit const mov case.
14483 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14484 Richard Sandiford <richard.sandiford@linaro.org>
14486 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14487 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14488 Add integer and movi cases.
14489 (movi-split-hf-df-sf split, fp16): New.
14490 (enabled): Added TARGET_FP_F16INST.
14491 * config/aarch64/iterators.md (GPF_HF): New.
14492 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14494 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14496 * config/aarch64/aarch64.c
14497 (aarch64_simd_container_mode): Add prototype.
14498 (aarch64_expand_mov_immediate): Add HI support.
14499 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14500 (aarch64_can_const_movi_rtx_p): New.
14501 (aarch64_preferred_reload_class):
14502 Remove restrictions of using FP registers for certain SIMD operations.
14503 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14504 (aarch64_valid_floating_const): Add integer move validation.
14505 (aarch64_simd_imm_scalar_p): Remove.
14506 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14507 (aarch64_legitimate_constant_p): Expand list of supported cases.
14508 * config/aarch64/aarch64-protos.h
14509 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14510 (aarch64_reinterpret_float_as_int): New.
14511 (aarch64_simd_imm_scalar_p): Remove.
14512 * config/aarch64/constraints.md (Uvi): New.
14513 (Dd): Split into Ds and new Dd.
14514 * config/aarch64/aarch64.md (*movsi_aarch64):
14516 (*movdi_aarch64): Add SIMD mov case.
14518 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14520 * tree-predcom.c: (struct chain): Handle store-store chain in which
14521 stores for elimination only store loop invariant values.
14522 (execute_pred_commoning_chain): Ditto.
14523 (prepare_initializers_chain_store_elim): Ditto.
14524 (prepare_finalizers): Ditto.
14525 (is_inv_store_elimination_chain): New function.
14526 (initialize_root_vars_store_elim_1): New function.
14528 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14530 * tree-predcom.c: Revise general description of the pass.
14531 (enum chain_type): New enum type for store elimination.
14532 (struct chain): New field supporting store elimination.
14533 (struct component): Ditto.
14534 (dump_chain): Dump store-stores chain.
14535 (release_chain): Release resources.
14536 (split_data_refs_to_components): Compute and create component
14537 contains only stores for elimination.
14538 (get_chain_last_ref_at): New function.
14539 (make_invariant_chain): Initialization.
14540 (make_rooted_chain): Specify chain type in parameter and record it.
14541 (add_looparound_copies): Skip for store-stores chain.
14542 (determine_roots_comp): Compute type of chain and pass it to
14544 (initialize_root_vars_store_elim_2): New function.
14545 (finalize_eliminated_stores): New function.
14546 (remove_stmt): Handle store for elimination.
14547 (execute_pred_commoning_chain): Execute predictive commoning on
14548 store-store chains.
14549 (determine_unroll_factor): Skip unroll for store-stores chain.
14550 (prepare_initializers_chain_store_elim): New function.
14551 (prepare_initializers_chain): Hanlde store-store chain.
14552 (prepare_finalizers_chain, prepare_finalizers): New function.
14553 (tree_predictive_commoning_loop): Return integer value indicating
14554 if loop is unrolled or lcssa form is corrupted.
14555 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14557 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14559 * tree-predcom.c (initialize_root): Delete.
14560 (execute_pred_commoning_chain): Initialize root vars and replace
14561 reference of non-combined chain directly, rather than call above
14564 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14566 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14567 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14569 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14571 * tree-predcom.c (struct chain): New field init_seq.
14572 (release_chain): Release init_seq.
14573 (prepare_initializers_chain): Record intialization stmts in above
14575 (insert_init_seqs): New function.
14576 (tree_predictive_commoning_loop): Call insert_init_seqs.
14578 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14580 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14582 2017-07-28 Richard Biener <rguenther@suse.de>
14584 * match.pd: Remove superfluous :c.
14585 * genmatch.c (simplify::id): Add member.
14586 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14588 (current_id): New global.
14589 (dt_node::parent): Move from ...
14590 (dt_operand::parent): ... here. Add for_id member.
14591 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14592 (decision_tree::find_node): Relax order requirement when
14593 merging DT_TRUE nodes to ones inbetween the current simplify
14594 and the one we try to merge with. Add diagnostic whenever
14595 we need to enforce pattern order by not merging.
14596 (decision_tree::insert): Set current_id.
14597 (decision_tree::print_node): Dump parent node and for_id.
14598 (parser::last_id): Add member.
14599 (parser::push_simplify): Assign unique id.
14600 (parser::parser): Initialize last_id.
14602 2017-07-28 Martin Liska <mliska@suse.cz>
14605 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14606 gimple_build_debug_bind.
14608 2017-07-28 Richard Biener <rguenther@suse.de>
14610 PR tree-optimization/81502
14611 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14613 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14615 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14616 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14617 for BIT_FIELD_REF args.
14618 * fold-const.c (make_bit_field_ref): Likewise.
14619 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14621 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14624 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14625 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14626 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14627 Or it into SANITIZER_UNDEFINED.
14628 * ubsan.c: Include gimple-fold.h and varasm.h.
14629 (ubsan_expand_ptr_ifn): New function.
14630 (instrument_pointer_overflow): New function.
14631 (maybe_instrument_pointer_overflow): New function.
14632 (instrument_object_size): Formatting fix.
14633 (pass_ubsan::execute): Call instrument_pointer_overflow
14634 and maybe_instrument_pointer_overflow.
14635 * internal-fn.c (expand_UBSAN_PTR): New function.
14636 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14637 * sanitizer.def (__ubsan_handle_pointer_overflow,
14638 __ubsan_handle_pointer_overflow_abort): New builtins.
14639 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14640 * internal-fn.def (UBSAN_PTR): New internal function.
14641 * opts.c (sanitizer_opts): Add pointer-overflow.
14642 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14643 * fold-const.c (build_range_check): Compute pointer range check in
14644 integral type if pointer arithmetics would be needed. Formatting
14647 2017-07-28 Martin Liska <mliska@suse.cz>
14650 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14651 parameters that are of a variable-length.
14653 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14655 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14657 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14658 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14659 (CRT_CALL_STATIC_FUNCTION): Likewise.
14660 (ASM_DEFAULT_SPEC): New define.
14661 (ASM_SPEC32): Likewise.
14662 (ASM_SPEC64): Likewise.
14663 (ASM_SPEC_COMMON): Likewise.
14664 (ASM_SPEC): Likewise.
14665 (INVALID_64BIT): Likewise.
14666 (LINK_OS_DEFAULT_SPEC): Likewise.
14667 (LINK_OS_SPEC32): Likewise.
14668 (LINK_OS_SPEC64): Likewise.
14669 (POWERPC_LINUX): Likewise.
14670 (PTRDIFF_TYPE): Likewise.
14671 (RESTORE_FP_PREFIX): Likewise.
14672 (RESTORE_FP_SUFFIX): Likewise.
14673 (SAVE_FP_PREFIX): Likewise.
14674 (SAVE_FP_SUFFIX): Likewise.
14675 (SIZE_TYPE): Likewise.
14676 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14677 (TARGET_64BIT): Likewise.
14678 (TARGET_64BIT): Likewise.
14679 (TARGET_AIX): Likewise.
14680 (WCHAR_TYPE_SIZE): Likewise.
14681 (WCHAR_TYPE): Undefine.
14682 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14683 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14684 (CPP_OS_RTEMS_SPEC): Delete.
14685 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14686 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14688 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14690 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14692 PR tree-optimization/81578
14693 * tree-parloops.c (build_new_reduction): Bail out if
14694 reduction_code isn't one of the standard OpenMP reductions.
14695 Move the details printing after that decision.
14697 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14699 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14700 related to reload_in_progress.
14701 (splat_input_operand): Likewise.
14702 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14704 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14706 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14707 (TARGET_INSTANTIATE_DECLS): Likewise.
14708 (legitimate_indexed_address_p): Delete reload_in_progress code.
14709 (rs6000_debug_legitimate_address_p): Likewise.
14710 (rs6000_eliminate_indexed_memrefs): Likewise.
14711 (rs6000_emit_le_vsx_store): Likewise.
14712 (rs6000_emit_move_si_sf_subreg): Likewise.
14713 (rs6000_emit_move): Likewise.
14714 (register_to_reg_type): Likewise.
14715 (rs6000_pre_atomic_barrier): Likewise.
14716 (rs6000_machopic_legitimize_pic_address): Likewise.
14717 (rs6000_allocate_stack_temp): Likewise.
14718 (rs6000_address_for_fpconvert): Likewise.
14719 (rs6000_address_for_altivec): Likewise.
14720 (rs6000_secondary_memory_needed_rtx): Delete function.
14721 (rs6000_check_sdmode): Likewise.
14722 (rs6000_alloc_sdmode_stack_slot): Likewise.
14723 (rs6000_instantiate_decls): Likewise.
14724 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14725 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14726 Delete reload_in_progress.
14727 (*vec_reload_and_plus_<mptrsize>): Likewise.
14728 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14729 (vsx_div_v2di): Likewise.
14730 (vsx_udiv_v2di): Likewise.
14732 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14734 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14735 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14736 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14737 (rs6000_debug_reg_global): Delete print of LRA status.
14738 (rs6000_option_override_internal): Delete dead LRA related code.
14739 (rs6000_lra_p): Delete function.
14740 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14742 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14744 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14745 * config/riscv/rtems.h: New file.
14747 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14748 Sudakshina Das <sudi.das@arm.com>
14750 * config/aarch64/aarch64.md
14751 (define_split for and<mode>3nr_compare): Move
14752 non aarch64_logical_operand to a register.
14753 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14754 register immediate operand to a register.
14755 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14757 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14759 PR middle-end/81564
14760 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14762 2017-07-27 Richard Biener <rguenther@suse.de>
14764 PR tree-optimization/81573
14765 PR tree-optimization/81494
14766 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14767 multi defuse cycle case.
14769 2017-07-27 Richard Biener <rguenther@suse.de>
14771 PR tree-optimization/81571
14772 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14775 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14777 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14778 earlier and only if MASK_FPU is set. Adjust formatting.
14780 2017-07-27 Martin Liska <mliska@suse.cz>
14782 * opt-functions.awk: Add validation of value of Init.
14783 * optc-gen.awk: Pass new argument.
14785 2017-07-27 Martin Liska <mliska@suse.cz>
14787 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14788 Fix wrong condition.
14790 2017-07-27 Martin Liska <mliska@suse.cz>
14792 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14793 BBs and edges seen by autoFDO.
14795 2017-07-27 Richard Biener <rguenther@suse.de>
14797 PR tree-optimization/81502
14798 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14799 with incompatible but same sized type.
14800 (execute_update_addresses_taken): Likewise.
14802 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14804 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14805 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14807 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14810 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14811 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14812 Change s_operand to memory_operand.
14814 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14816 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14817 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14818 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14819 Emit instructions rather than returning an expression. Handle TFmode
14820 and KFmode by casting to TImode.
14821 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14822 (rs6000_emit_le_vsx_store): Likewise.
14823 * config/rs6000/vsx.md (VSX_TI): New iterator.
14824 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14825 (*vsx_le_undo_permute_<mode>): Likewise.
14826 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14827 emit the split sequence.
14828 (*vsx_le_perm_store_<mode>): Likewise.
14830 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14832 PR tree-optimization/81555
14833 PR tree-optimization/81556
14834 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14835 if true, force CHANGED for the recursive invocation.
14836 (reassociate_bb): Remember original length of ops array, pass
14837 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14839 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14840 attributes for noipa attribute. For naked attribute use
14841 lookup_attribute first before lookup_attribute_spec.
14842 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14844 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14845 for functions with noipa attribute.
14846 (cgraph_externally_visible_p): Return true for functions with noipa
14848 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14849 for functions with noipa attribute.
14850 * doc/extend.texi: Document noipa function attribute.
14851 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14852 also for functions with noipa attribute.
14853 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14855 2017-07-26 Andrew Pinski <apinski@cavium.com>
14857 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14858 vec_unalign_load_cost and vec_unalign_store_cost.
14860 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14862 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14863 -mvsx-small-integer option.
14864 (ISA_3_0_MASKS_IEEE): Likewise.
14865 (OTHER_VSX_VECTOR_MASKS): Likewise.
14866 (POWERPC_MASKS): Likewise.
14867 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14868 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14869 code, only testing for DImode being allowed in non-VSX floating
14871 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14872 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14874 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14875 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14876 TARGET_P8_VECTOR test.
14877 (rs6000_secondary_reload_simple_move): Likewise.
14878 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14879 since TARGET_P9_VECTOR was already tested.
14880 (rs6000_opt_masks): Remove -mvsx-small-integer.
14881 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14882 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14884 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14885 test for TARGET_VEXTRACTUB was used, and that uses
14887 (p9 extract splitter): Likewise.
14888 (vsx_extract_<mode>_di_p9): Likewise.
14889 (vsx_extract_<mode>_store_p9): Likewise.
14890 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14891 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14892 the elimination of TARGET_VSX_SMALL_INTEGER.
14893 (vsx_extract_<mode>_p8): Likewise.
14894 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14895 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14896 (vsx_set_<mode>_p9): Likewise.
14897 (vsx_set_v4sf_p9): Likewise.
14898 (vsx_set_v4sf_p9_zero): Likewise.
14899 (vsx_insert_extract_v4sf_p9): Likewise.
14900 (vsx_insert_extract_v4sf_p9_2): Likewise.
14901 * config/rs6000/rs6000.md (sign extend splitter): Change
14902 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14903 (floatsi<mode>2_lfiwax_mem): Likewise.
14904 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14905 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14906 since a test for TARGET_P9_VECTOR was used.
14907 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14908 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14909 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14910 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14911 TARGET_P8_VECTOR test.
14912 (fix_trunc<mode>si2_stfiwx): Likewise.
14913 (fix_trunc<mode>si2_internal): Likewise.
14914 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14915 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14917 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14918 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14919 TARGET_P8_VECTOR test.
14920 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14921 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14922 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14924 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14925 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14926 since a test for TARGET_P9_VECTOR was used.
14927 (splitter for loading small constants): Likewise.
14929 2017-07-26 Andrew Pinski <apinski@cavium.com>
14931 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14934 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14937 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14938 (fp_valid_at): Likewise.
14940 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14942 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14943 (qdf24xx_addrcost_table): Likewise.
14944 (cortexa57_tunings): Update to use generic_branch_cost.
14945 (cortexa72_tunings): Likewise.
14946 (cortexa73_tunings): Likewise.
14947 (qdf24xx_tunings): Likewise.
14949 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14951 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14952 (thunderx2t99_branch_cost): Likewise.
14953 (cortexa35_tunings): Update to use generic_branch_cost.
14954 (cortexa53_tunings): Likewise.
14955 (cortexa57_tunings): Likewise.
14956 (cortexa72_tunings): Likewise.
14957 (cortexa73_tunings): Likewise.
14958 (thunderx2t99_tunings): Likewise.
14960 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14962 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14963 (sparc_option_override): Honour MASK_FSMULD.
14964 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14965 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14966 * config/sparc/sparc.opt (mfsmuld): New option.
14967 * doc/invoke.texi (mfsmuld): Document option.
14969 2017-07-26 Marek Polacek <polacek@redhat.com>
14971 PR middle-end/70992
14972 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14974 2017-07-26 Richard Biener <rguenther@suse.de>
14976 * gimple-match-head.c (do_valueize): Return OP if valueize
14978 (get_def): New helper to get at the def stmt of a SSA name
14979 if valueize allows.
14980 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14981 do_valueize to get at the def stmt.
14982 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14984 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14986 PR middle-end/46932
14987 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14989 2017-07-26 Martin Liska <mliska@suse.cz>
14992 * function.c (expand_function_start): Make expansion of
14993 nonlocal_goto_save_area after parm_birth_insn.
14995 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14997 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14998 from all CPU target flags enable members.
15000 2017-07-26 Richard Biener <rguenther@suse.de>
15002 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15003 (decision_tree::gen): Make 'type' const.
15004 (write_predicate): Likewise.
15006 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15008 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15009 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15010 (rs6000_option_override_internal): Likewise.
15011 (rs6000_expand_vector_set): Likewise.
15012 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15013 (TARGET_UPPER_REGS_SF): Likewise.
15014 (TARGET_UPPER_REGS_DI): Likewise.
15015 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15016 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15017 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15018 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15019 (Splitters for DI constants in Altivec registers): Likewise.
15020 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15021 (vsx_set_v4sf_p9): Likewise.
15022 (vsx_set_v4sf_p9_zero): Likewise.
15023 (vsx_insert_extract_v4sf_p9): Likewise.
15024 (vsx_insert_extract_v4sf_p9_2): Likewise.
15026 2017-07-25 Carl Love <cel@us.ibm.com>
15028 * doc/extend.texi: Update the built-in documentation file for the
15029 existing built-in functions
15030 vector signed char vec_cnttz (vector signed char);
15031 vector unsigned char vec_cnttz (vector unsigned char);
15032 vector signed short vec_cnttz (vector signed short);
15033 vector unsigned short vec_cnttz (vector unsigned short);
15034 vector signed int vec_cnttz (vector signed int);
15035 vector unsigned int vec_cnttz (vector unsigned int);
15036 vector signed long long vec_cnttz (vector signed long long);
15037 vector unsigned long long vec_cnttz (vector unsigned long long);
15039 2017-07-25 Andrew Pinski <apinski@cavium.com>
15041 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15042 accesses where the use is for the first operand of a BIT_INSERT.
15044 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15047 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15048 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15050 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15052 * config/i386/gstabs.h: Delete.
15053 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15055 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15057 * config/i386/i386.c (ix86_decompose_address): Do not check for
15058 register RTX when looking at index_reg or base_reg.
15059 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15061 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15063 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15064 to update EH info here.
15066 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15068 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15070 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15072 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15074 2017-07-25 Torsten Duwe <duwe@suse.de>
15076 * common.opt: Introduce -fpatchable-function-entry
15077 command line option, and its variables function_entry_patch_area_size
15078 and function_entry_patch_area_start.
15079 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15080 including a two-value parser.
15081 * target.def (print_patchable_function_entry): New target hook.
15082 * targhooks.h (default_print_patchable_function_entry): New function.
15083 * targhooks.c (default_print_patchable_function_entry): Likewise.
15084 * toplev.c (process_options): Switch off IPA-RA if
15085 patchable function entries are being generated.
15086 * varasm.c (assemble_start_function): Look at the
15087 patchable-function-entry command line switch and current
15088 function attributes and maybe generate NOP instructions by
15089 calling the print_patchable_function_entry hook.
15090 * doc/extend.texi: Document patchable_function_entry attribute.
15091 * doc/invoke.texi: Document -fpatchable_function_entry
15092 command line option.
15093 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15095 * doc/tm.texi: Re-generate.
15097 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15100 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15101 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15103 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15105 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15107 2017-07-25 Richard Biener <rguenther@suse.de>
15109 PR tree-optimization/81455
15110 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15111 not walk in cycles when looking for guards.
15113 2017-07-25 Richard Biener <rguenther@suse.de>
15115 PR tree-optimization/81529
15116 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15117 when optimizing backedge uses.
15119 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15121 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15123 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15124 to dl_section_ref. On AIX, append an expression to subtract
15125 the size of the section length to dl_section_ref.
15127 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15129 * configure.ac: If any of the config.* scripts fail, exit 1.
15130 * configure: Regenerate.
15132 2017-07-25 Richard Biener <rguenther@suse.de>
15134 PR middle-end/81546
15135 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15136 of immediate uses, be more verbose on errors.
15138 2017-07-25 Richard Biener <rguenther@suse.de>
15140 PR tree-optimization/81510
15141 * tree-vect-loop.c (vect_is_simple_reduction): When the
15142 reduction stmt is not inside the loop bail out.
15144 2017-07-25 Richard Biener <rguenther@suse.de>
15146 PR tree-optimization/81303
15147 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15148 profitability check against LOOP_VINFO_NITERSM1.
15150 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15152 * domwalk.c (cmp_bb_postorder): Simplify.
15153 (sort_bbs_postorder): New function. Use it...
15154 (dom_walker::walk): ...here to optimize common cases.
15156 2017-07-25 Martin Liska <mliska@suse.cz>
15159 * ipa-visibility.c (function_and_variable_visibility): Make the
15160 redirection just on target that supports aliasing.
15161 Fix GNU coding style.
15163 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15166 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15168 * config/arm/rtems.h: Likewise.
15169 * config/bfin/rtems.h: Likewise.
15170 * config/i386/rtemself.h: Likewise.
15171 * config/lm32/rtems.h: Likewise.
15172 * config/m32c/rtems.h: Likewise.
15173 * config/m68k/rtemself.h: Likewise.
15174 * config/microblaze/rtems.h: Likewise.
15175 * config/mips/rtems.h: Likewise.
15176 * config/moxie/rtems.h: Likewise.
15177 * config/nios2/rtems.h: Likewise.
15178 * config/powerpcspe/rtems.h: Likewise.
15179 * config/rs6000/rtems.h: Likewise.
15180 * config/rtems.h: Likewise.
15181 * config/sh/rtems.h: Likewise.
15182 * config/sh/rtemself.h: Likewise.
15183 * config/sparc/rtemself.h: Likewise.
15185 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15188 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15189 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15190 * tree-ssa-structalias.c (alias_get_name): Same.
15192 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15195 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15196 instructions if no du chain is found.
15198 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15200 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15202 2017-07-25 Richard Biener <rguenther@suse.de>
15204 PR middle-end/81505
15205 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15208 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15210 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15211 upper-regs options.
15212 (ISA_2_7_MASKS_SERVER): Likewise.
15213 (ISA_3_0_MASKS_IEEE): Likewise.
15214 (OTHER_P8_VECTOR_MASKS): Likewise.
15215 (OTHER_VSX_VECTOR_MASKS): Likewise.
15216 (POWERPC_MASKS): Likewise.
15217 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15218 duplicate list of options.
15219 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15220 explicit -mupper-regs options.
15221 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15222 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15223 alias for -mupper-regs-df.
15224 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15225 (rs6000_init_hard_regno_mode_ok): Likewise.
15226 (rs6000_option_override_internal): Likewise.
15227 (rs6000_opt_masks): Likewise.
15228 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15229 options in terms of whether -mvsx or -mpower8-vector was used.
15230 (TARGET_UPPER_REGS_DI): Likewise.
15231 (TARGET_UPPER_REGS_SF): Likewise.
15232 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15233 -mupper-regs-* options.
15235 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15237 * passes.c (emergency_dump_function): Print some empty lines and a
15238 header before the RTL dump.
15240 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15242 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15244 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15247 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15248 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15250 2017-07-24 Carl Love <cel@us.ibm.com>
15252 * config/rs6000/rs6000-c.c: Add support for built-in functions
15253 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15254 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15255 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15256 vec_extract_fp_from_shortl): Add defines for the two builtins.
15257 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15258 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15260 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15261 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15262 * doc/extend.texi: Update the built-in documentation file for the
15263 new built-in function.
15265 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15268 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15270 * doc/generic.texi: Likewise.
15271 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15272 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15274 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15276 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15277 (aarch64_mls_elt_merge<mode>): Likewise.
15279 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15281 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15282 having __cxa_atexit.
15284 2017-07-23 Michael Collison <michael.collison@arm.com>
15286 * config/arm/arm.c (arm_option_override): Deprecate
15287 use of -mstructure-size-boundary.
15288 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15289 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15291 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15294 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15295 Reduce cost estimate for direct moves.
15297 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15300 * config/i386/i386.c (ix86_option_override_internal): Disable
15301 BMI, BMI2 and TBM instructions for -m16.
15303 2017-07-21 Carl Love <cel@us.ibm.com>
15305 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15306 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15307 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15308 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15309 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15310 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15311 VMULOSW): New enum "unspec" values.
15312 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15313 altivec_vmulosw): New patterns.
15314 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15315 VMULOSW): Add definitions.
15317 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15319 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15320 (qdf24xx): Likewise.
15321 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15322 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15323 (AARCH64_FL_V8_1): Renumber.
15324 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15325 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15326 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15327 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15328 rdma to feature modifiers list.
15330 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15332 PR middle-end/56727
15333 * ipa-visibility (function_and_variable_visibility): Convert
15334 recursive PLT call to direct call if appropriate.
15336 2017-07-21 Andrew Pinski <apinski@cavium.com>
15338 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15339 operand 1 to see if the types precision matches.
15340 * fold-const.c (operand_equal_p): Likewise.
15342 2017-07-21 Richard Biener <rguenther@suse.de>
15344 PR tree-optimization/81303
15345 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15346 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15347 (vect_peeling_hash_get_lowest_cost): Adjust.
15348 (vect_enhance_data_refs_alignment): Likewise. Use
15349 vect_get_peeling_costs_all_drs to compute the penalty for no
15350 peeling to match up costs.
15352 2017-07-21 Richard Biener <rguenther@suse.de>
15354 PR tree-optimization/81500
15355 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15356 we didn't identify a reduction path.
15358 2017-07-21 Tom de Vries <tom@codesourcery.com>
15359 Cesar Philippidis <cesar@codesourcery.com>
15361 PR gcov-profile/81442
15362 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15365 2017-07-21 Tom de Vries <tom@codesourcery.com>
15368 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15370 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15371 nvptx_override_options_after_change.
15373 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15375 * dwarf2out.c (output_file_names): Avoid double testing for
15376 dwarf_version >= 5.
15378 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15380 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15382 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15384 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15386 (try_crossjump_to_edge): Do not punt on partitioned functions.
15388 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15390 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15391 Put all BBs reachable only via paths crossing cold region to cold
15393 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15395 2016-07-21 Richard Biener <rguenther@suse.de>
15397 PR tree-optimization/81303
15398 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15399 into account prologue and epilogue iterations when raising
15400 min_profitable_iters to sth at least covering one vector iteration.
15402 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15404 * config/arm/arm.c (arm_test_cpu_arch_dat):
15407 2017-07-20 Nathan Sidwell <nathan@acm.org>
15409 Remove TYPE_METHODS.
15410 * tree.h (TYPE_METHODS): Delete.
15411 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15412 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15413 (dbxout_type_methods): Scan TYPE_FIELDS.
15414 (dbxout_type): Don't check TYPE_METHODS here.
15415 * function.c (use_register_for_decl): Always ignore register for
15416 class types when not optimizing.
15417 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15418 * tree.c (free_lang_data_in_type): Stitch out member functions and
15419 templates from TYPE_FIELDS.
15420 (build_distinct_type_copy, verify_type_variant,
15421 verify_type): Member fns are on TYPE_FIELDS.
15422 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15423 * tree-pretty-print.c (dump_generic_node): Likewise.
15425 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15428 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15429 V2TImode and V4TImode.
15430 (ix86_expand_vector_extract): Likewise.
15431 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15432 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15433 (ssescalarmode): Handle V4TImode and V2TImode.
15434 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15435 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15436 (VEXTRACTI128_MODE): New mode iterator.
15437 (splitter for *vec_extractv?ti first element): New.
15438 (VEC_INIT_MODE): New mode iterator.
15439 (vec_init<mode>): Consolidate 3 expanders into one using
15440 VEC_INIT_MODE mode iterator.
15442 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15444 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15445 non_spilled_static_chain_regno_p.
15447 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15449 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15451 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15453 * bb-reorder.c (connect_traces): Allow copying of blocks within
15456 2017-07-20 Richard Biener <rguenther@suse.de>
15458 * gimple.h (gimple_phi_result): Add gphi * overload.
15459 (gimple_phi_result_ptr): Likewise.
15460 (gimple_phi_arg): Likewise. Adjust index assert to only
15461 allow actual argument accesses rather than all slots available
15463 (gimple_phi_arg_def): Add gphi * overload.
15464 * tree-phinodes.c (make_phi_node): Initialize only actual
15466 (resize_phi_node): Clear memory not covered by old node,
15467 do not initialize excess argument slots.
15468 (reserve_phi_args_for_new_edge): Initialize new argument slot
15471 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15473 PR tree-optimization/81388
15475 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15477 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15478 by removing computation of may_be_zero.
15480 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15481 Tom de Vries <tom@codesourcery.com>
15483 PR middle-end/81030
15484 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15485 when gimple level profile disagrees with what RTL expander did.
15487 2017-07-20 Richard Biener <rguenther@suse.de>
15489 PR tree-optimization/61171
15490 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15491 (vect_analyze_stmt): Add slp instance parameter.
15492 (vectorizable_reduction): Likewise.
15493 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15494 (vect_is_simple_reduction): Deal with chains not detected
15495 as SLP reduction chain, specifically not properly associated
15496 chains containing a mix of plus/minus.
15497 (get_reduction_op): Remove.
15498 (get_initial_defs_for_reduction): Simplify, pass in whether
15499 this is a reduction chain, pass in the SLP node for the PHIs.
15500 (vect_create_epilog_for_reduction): Get the SLP instance as
15502 (vectorizable_reduction): Get the SLP instance as arg.
15503 During analysis remember the SLP node with the PHIs in the
15504 instance. Simplify getting at the vectorized reduction PHIs.
15505 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15506 through SLP instance.
15507 (vect_slp_analyze_operations): Likewise.
15508 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15509 (vect_transform_stmt): Likewise.
15511 2017-07-20 Tom de Vries <tom@codesourcery.com>
15513 PR tree-optimization/81489
15514 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15515 read of phi arg location to before loop that modifies phi.
15517 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15519 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15522 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15524 PR middle-end/81331
15525 * except.c (execute): Fix ordering issue.
15527 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15529 PR rtl-optimization/81423
15530 * combine.c (make_compound_operation_int): Don't try to optimize
15531 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15533 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15535 PR rtl-optimization/81423
15536 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15537 with a constant that is -1 in the truncated to mode.
15539 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15541 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15542 (determine_unlikely_bbs): ... here.
15543 * predict.h (propagate_unlikely_bbs_forward): Declare.
15544 * cfgexpand.c (pass_expand::execute): Use it.
15545 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15547 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15548 propagate_unlikely_bbs_forward.
15550 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15552 PR middle-end/81331
15553 * except.c (maybe_add_nop_after_section_switch): New function.
15556 2017-07-19 Tom de Vries <tom@codesourcery.com>
15558 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15560 2017-07-19 Tom de Vries <tom@codesourcery.com>
15562 * gimple.h (gimple_phi_arg): Make assert more strict.
15564 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15566 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15567 * config/rs6000/mmintrin.h: New file.
15568 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15570 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15572 PR tree-optimization/81346
15573 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15575 2017-07-19 Tom de Vries <tom@codesourcery.com>
15577 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15579 2017-07-19 Tom de Vries <tom@codesourcery.com>
15581 * config/nvptx/nvptx-modes.def: Add V2DImode.
15582 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15583 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15584 (nvptx_output_mov_insn): Handle lack of mov.b128.
15585 (nvptx_print_operand): Handle 'H' and 'L' codes.
15586 (nvptx_vector_mode_supported): Allow V2DImode.
15587 (nvptx_preferred_simd_mode): New function.
15588 (nvptx_data_alignment): New function.
15589 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15590 nvptx_preferred_simd_mode.
15591 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15593 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15595 2017-07-19 Tom de Vries <tom@codesourcery.com>
15597 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15598 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15599 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15600 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15601 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15602 (mov<VECIM>_insn): New define_insn.
15603 (define_expand "mov<VECIM>): New define_expand.
15605 2017-07-19 Tom de Vries <tom@codesourcery.com>
15607 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15609 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15611 PR tree-optimization/81346
15612 * fold-const.h (fold_div_compare, range_check_type): Declare.
15613 * fold-const.c (range_check_type): New function.
15614 (build_range_check): Use range_check_type.
15615 (fold_div_compare): No longer static, rewritten into
15616 a match.pd helper function.
15617 (fold_comparison): Don't call fold_div_compare here.
15618 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15619 as helper function.
15621 2017-07-19 Nathan Sidwell <nathan@acm.org>
15623 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15624 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15625 * tree.c (find_decls_types_r, verify_type): Use
15626 TYPE_{MIN,MAX}_VALUE_RAW.
15627 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15628 (hash_tree): Likewise.
15629 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15631 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15634 2017-07-18 Tom de Vries <tom@codesourcery.com>
15636 PR middle-end/81464
15637 * omp-expand.c (expand_omp_for_static_chunk): Handle
15638 equal-argument loop exit phi.
15640 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15643 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15644 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15645 operand 2 predicate.
15646 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15647 operand 2 predicate.
15648 (ror,rol -> rorx splitters): Use const_int_operand as
15649 operand 2 predicate.
15651 2017-06-18 Richard Biener <rguenther@suse.de>
15653 PR tree-optimization/81410
15654 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15655 the gap in the ! slp_perm SLP case after each group.
15657 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15659 PR middle-end/81463
15660 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15663 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15665 PR middle-end/81462
15666 * predict.c (set_even_probabilities): Cleanup; do not affect
15667 probabilities that are already known.
15668 (combine_predictions_for_bb): Call even when count is set.
15670 2017-07-18 Nathan Sidwell <nathan@acm.org>
15672 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15675 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15678 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15679 optimization for loop niter analysis.
15681 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15684 * config/avr/avr.c (avr_optimize_casesi): Don't use
15685 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15687 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15689 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15690 body_cost_vec from _vect_peel_extended_info.
15691 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15692 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15695 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15697 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15699 2017-07-18 Richard Biener <rguenther@suse.de>
15701 PR tree-optimization/80620
15702 PR tree-optimization/81403
15703 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15704 info when re-using a VN table entry.
15706 2017-07-18 Richard Biener <rguenther@suse.de>
15708 PR tree-optimization/81418
15709 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15710 vectype_in. Verify that with lane-reducing reduction operations
15711 we have a single def-use cycle.
15713 2017-07-17 Carl Love <cel@us.ibm.com>
15715 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15717 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15718 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15719 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15721 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15722 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15723 VMULOSW): New enum "unspec" values.
15724 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15725 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15726 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15727 altivec_vmulosw): New patterns.
15728 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15729 VMULOSW): Add definitions.
15730 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15732 * config/alpha/alpha.c: Include predict.h.
15734 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15736 * tree-vrp.c (compare_assert_loc): Fix comparison function
15737 to return predictable results.
15739 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15741 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15743 (subdi3): Likewise.
15744 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15745 * doc/invoke.texi (mexpand-adddi): Update text.
15747 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15749 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15750 that also clobbers the CC register. The old expand code is moved
15752 (*arc_clzsi2): ... here.
15753 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15754 the CC register. The old expand code is moved to ...
15755 (arc_ctzsi2): ... here.
15757 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15759 * config/arc/arc.opt (mindexed-loads): Use initial value
15760 TARGET_INDEXED_LOADS_DEFAULT.
15761 (mauto-modify-reg): Use initial value
15762 TARGET_AUTO_MODIFY_REG_DEFAULT.
15763 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15764 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15765 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15766 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15768 2017-07-17 Martin Liska <mliska@suse.cz>
15771 * opts.c (finish_options): Do not allow -fgnu-tm
15772 w/ -fsanitize={kernel-,}address. Say sorry.
15774 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15777 * tree-loop-distribution.c (classify_partition): Only assert on
15778 numer of iterations.
15779 (merge_dep_scc_partitions): Delete prameter. Update function call.
15780 (distribute_loop): Remove code handling loop with unknown niters.
15781 (pass_loop_distribution::execute): Skip loop with unknown niters.
15783 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15786 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15787 function sort_partitions_by_post_order.
15789 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15791 PR tree-optimization/81374
15792 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15793 the max index of basic blocks, rather than number of basic blocks.
15795 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15797 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15799 (arc_legitimate_pic_operand_p): Likewise.
15800 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15802 (arc_needs_pcl_p): Likewise.
15803 (arc_legitimate_pc_offset_p): Likewise.
15804 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15805 function is also used in constrains.md.
15806 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15807 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15808 PLUS. Only return true/false in known cases, otherwise assert.
15809 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15810 is already called in arc_legitimate_constant_p.
15811 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15813 (LEGITIMATE_PIC_OPERAND_P): Use
15814 arc_raw_symbolic_reference_mentioned_p function.
15815 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15820 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15821 Andrew Burgess <andrew.burgess@embecosm.com>
15823 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15824 (arc_return_address_register): New function.
15825 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15826 (arc_handle_fndecl_attribute): Add naked attribute.
15827 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15828 (TARGET_WARN_FUNC_RETURN): Likewise.
15829 (arc_allocate_stack_slots_for_args): New function.
15830 (arc_warn_func_return): Likewise.
15831 (machine_function): Change type fn_type.
15832 (arc_compute_function_type): Consider new naked function type,
15833 change function return type.
15834 (arc_must_save_register): Adapt to handle new
15835 arc_compute_function_type's return type.
15836 (arc_expand_prologue): Likewise.
15837 (arc_expand_epilogue): Likewise.
15838 (arc_return_address_regs): Delete.
15839 (arc_return_address_register): New function.
15840 (arc_epilogue_uses): Use above function.
15841 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15842 (arc_function_type): Change encoding, add naked type.
15843 (ARC_INTERRUPT_P): Change to handle the new encoding.
15844 (ARC_FAST_INTERRUPT_P): Likewise.
15845 (ARC_NORMAL_P): Define.
15846 (ARC_NAKED_P): Likewise.
15847 (arc_compute_function_type): Delete prototype.
15848 * config/arc/arc.md (in_ret_delay_slot): Use
15849 arc_return_address_register function.
15850 (simple_return): Likewise.
15851 (p_return_i): Likewise.
15853 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15855 PR tree-optimization/81428
15856 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15857 can't be built for those types.
15859 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15861 Remove stuff dead since r239246.
15863 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15864 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15865 (avr_inform_devices): Remove dead stuff.
15867 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15869 * config/arm/arm_neon.h: Fix softp typo.
15871 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15873 PR tree-optimization/81365
15874 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15875 aggregate moves onto bb predecessor edges, make sure there are no
15876 loads that could alias the lhs in between the start of bb and the
15879 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15882 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15883 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15884 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15886 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15888 PR tree-optimization/81396
15889 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15890 (init_symbolic_number): Initialize it to 1.
15891 (perform_symbolic_merge): Add n_ops from both operands into the new
15893 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15894 without base_addr as useless if they need more than one operation.
15895 (bswap_replace): Handle !bswap case for NULL base_addr.
15897 2017-07-17 Tom de Vries <tom@codesourcery.com>
15900 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15903 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15905 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15906 conditional builtin define __FIX_LEON3FT_B2BST.
15908 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15910 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15911 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15914 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15916 PR rtl-optimization/81424
15917 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15918 to remove potential trapping from operands if -fnon-call-exceptions.
15920 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15922 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15923 profile_proability for scalling.
15924 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15926 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15928 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15930 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15932 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15933 fixpoint arithmetics.
15935 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15937 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15938 fixpoint arithmetics.
15940 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15942 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15943 fixpoint arithmetics.
15945 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15947 * profile-count.h (profile_probability::from_reg_br_prob_note,
15948 profile_probability::to_reg_br_prob_note): New functions.
15949 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15950 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15951 * predict.c (probability_reliable_p): Update.
15952 (edge_probability_reliable_p): Update.
15953 (br_prob_note_reliable_p): Update.
15954 (invert_br_probabilities): Update.
15955 (add_reg_br_prob_note): New function.
15956 (combine_predictions_for_insn): Update.
15957 * asan.c (asan_clear_shadow): Update.
15958 * cfgbuild.c (compute_outgoing_frequencies): Update.
15959 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15960 (update_br_prob_note): Update.
15961 (rtl_verify_edges): Update.
15962 (purge_dead_edges): Update.
15963 (fixup_reorder_chain): Update.
15964 * emit-rtl.c (try_split): Update.
15965 * ifcvt.c (cond_exec_process_insns): Update.
15966 (cond_exec_process_if_block): Update.
15967 (dead_or_predicable): Update.
15968 * internal-fn.c (expand_addsub_overflow): Update.
15969 (expand_neg_overflow): Update.
15970 (expand_mul_overflow): Update.
15971 * loop-doloop.c (doloop_modify): Update.
15972 * loop-unroll.c (compare_and_jump_seq): Update.
15973 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15974 * predict.h: Update.
15975 * reorg.c (mostly_true_jump): Update.
15977 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15978 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15979 * config/arc/arc.c: (emit_unlikely_jump): Update.
15980 * config/arm/arm.c: (emit_unlikely_jump): Update.
15981 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15982 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15983 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15984 (ix86_print_operand): Update.
15985 (ix86_split_fp_branch): Update.
15986 (predict_jump): Update.
15987 * config/ia64/ia64.c (ia64_print_operand): Update.
15988 * config/mmix/mmix.c (mmix_print_operand): Update.
15989 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15990 (rs6000_expand_split_stack_prologue): Update.
15991 * config/rs6000/rs6000.c: Update.
15992 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15993 (s390_expand_vec_movstr): Update.
15994 (s390_expand_cs_tdsi): Update.
15995 (s390_expand_split_stack_prologue): Update.
15996 * config/sh/sh.c (sh_print_operand): Update.
15997 (expand_cbranchsi4): Update.
15998 (expand_cbranchdi4): Update.
15999 * config/sparc/sparc.c (output_v9branch): Update.
16000 * config/spu/spu.c (get_branch_target): Update.
16001 (ea_load_store_inline): Update.
16002 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16003 * config/tilepro/tilepro.c: Update.
16005 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16007 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16008 (gimplify_save_expr): Likewise.
16010 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16012 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16014 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16016 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16018 * timevar.def (TV_IPA_FNSUMMARY): Define.
16020 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16022 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16023 to back store errata sensitive sequence from being generated.
16024 (sqrtdf2_fix): Likewise.
16026 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16028 * tree-ssa-threadupdate.c (compute_path_counts,
16029 update_joiner_offpath_counts): Use profile_probability.
16031 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16034 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16036 * config/arm/arm-c.c (arm_cpu_builtins): Define
16037 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16039 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16041 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16042 array entries to represent __ieee128 versions of the
16043 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16044 scalar_extract_sig, and scalar_insert_exp built-in functions.
16045 (altivec_resolve_overloaded_builtin): Add special case handling
16046 for the __builtin_scalar_insert_exp function, as represented by
16047 the P9V_BUILTIN_VEC_VSIEDP constant.
16048 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16049 exponent support for __ieee128 argument.
16050 (VSESQP): Add scalar extract signature support for __ieee128
16052 (VSTDCNQP): Add scalar test negative support for __ieee128
16054 (VSIEQP): Add scalar insert exponent support for __int128 argument
16055 with __ieee128 result.
16056 (VSIEQPF): Add scalar insert exponent support for __ieee128
16057 argument with __ieee128 result.
16058 (VSTDCQP): Add scalar test data class support for __ieee128
16060 (VSTDCNQP): Add overload support for scalar test negative with
16061 __ieee128 argument.
16062 (VSTDCQP): Add overload support for scalar test data class
16063 __ieee128 argument.
16064 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16065 UNSPEC_VSX_SXSIGDP.
16066 (UNSPEC_VSX_SIEXPQP): New constant.
16067 (xsxexpqp): New insn for VSX scalar extract exponent quad
16069 (xsxsigqp): New insn for VSX scalar extract significand quad
16071 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16072 precision with floating point argument.
16073 (xststdcqp): New expand for VSX scalar test data class quad
16075 (xststdcnegqp): New expand for VSX scalar test negative quad
16077 (xststdcqp): New insn to match expansions for VSX scalar test data
16078 class quad precision and VSX scalar test negative quad precision.
16079 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16080 special case operand checking to enforce that second operand of
16081 VSX scalar test data class with quad precision argument is a 7-bit
16083 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16084 prototypes and descriptions of __ieee128 versions of
16085 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16086 scalar_test_data_class, and scalar_test_neg built-in functions.
16088 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16090 PR tree-optimization/81162
16091 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16092 replace a negate with an add.
16094 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16096 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16098 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16100 * config/arm/arm-c.c (arm_cpu_builtins): Define
16101 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16103 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16105 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16106 (armv8-r): Set ARM Cortex-R52 as default CPU.
16107 * config/arm/arm-tables.opt: Regenerate.
16108 * config/arm/arm-tune.md: Regenerate.
16109 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16111 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16112 extension for -mcpu=cortex-r52.
16114 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16116 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16117 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16118 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16119 (fp-armv8): Define it as FP_ARMv8 only.
16120 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16121 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16123 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16124 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16125 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16126 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16127 than TARGET_FPU_ARMV8.
16128 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16129 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16130 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16132 * config/arm/neon.md (neon_vrint): Likewise.
16133 (neon_vcvt): Likewise.
16134 (neon_<fmaxmin_op><mode>): Likewise.
16135 (<fmaxmin><mode>3): Likewise.
16136 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16137 * config/arm/predicates.md (arm_cond_move_operator): Check against
16138 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16140 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16142 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16143 to top of function.
16145 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16147 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16148 loop in comment with memset.
16150 2017-07-14 Martin Liska <mliska@suse.cz>
16152 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16153 * dwarf2out.c (is_java): Remove the function.
16154 (output_pubname): Remove usage of the function.
16155 (lower_bound_default): Remove usage of DW_LANG_Java.
16156 (gen_compile_unit_die): Likewise.
16157 * gcc.c: Remove compiler defaults for .java and .zip files.
16158 * gimple-expr.c (remove_suffix): Change as there's no longer
16159 extension than 4-letter one.
16160 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16161 (gimplify_save_expr): Likewise.
16162 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16163 as it's possible even for other languages than Java.
16164 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16165 * lto-opts.c (lto_write_options): Remove reference to Java.
16166 * opts.c (strip_off_ending): Update file extension handling.
16167 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16168 * tree-eh.c (lower_resx): Likewise.
16169 * tree.c (free_lang_data_in_type): Remove dead code.
16170 (find_decls_types_r): Likewise.
16171 (build_common_builtin_nodes): Remove Java from a comment.
16172 (verify_type): Remove dead code.
16173 * varasm.c (assemble_external): Remove Java from a comment.
16175 2017-07-14 Martin Liska <mliska@suse.cz>
16177 * opts.c (finish_options): Add quotes.
16178 (common_handle_option): Likewise.
16180 2017-07-14 Martin Liska <mliska@suse.cz>
16182 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16183 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16184 Remove N_SO_PASCAL.
16185 * dwarf2out.c (lower_bound_default): Do not handle
16187 (gen_compile_unit_die): Likewise.
16188 * gcc.c: Remove default extension binding for GNU Pascal.
16189 * stmt.c: Remove Pascal language from a comment.
16190 * xcoffout.c: Likewise.
16192 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16195 * diagnostic-show-locus.c (fixit_cmp): New function.
16196 (layout::layout): Sort m_fixit_hints.
16197 (column_range::column_range): Assert that the values are valid.
16198 (struct char_span): New struct.
16199 (correction::overwrite): New method.
16200 (struct source_line): New struct.
16201 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16202 calls in terms of classes source_line and char_span, and
16203 correction::overwrite.
16204 (selftest::test_overlapped_fixit_printing_2): New function.
16205 (selftest::diagnostic_show_locus_c_tests): Call it.
16207 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16209 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16210 early if there is no lhs.
16212 2017-07-13 Martin Liska <mliska@suse.cz>
16214 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16215 (gen_reference_type_die): Likewise.
16216 * stor-layout.c: Remove Pascal-related comment.
16218 2017-07-13 Martin Liska <mliska@suse.cz>
16220 * opts.c (finish_options): Add quotes to error messages.
16221 (parse_sanitizer_options): Likewise.
16223 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16225 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16227 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16229 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16231 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16233 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16235 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16237 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16240 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16241 provides the hardware capability bits, define the macro
16242 __BUILTIN_CPU_SUPPORTS__.
16243 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16244 if GLIBC does not provide the hardware capability bits. Add a
16245 gcc_unreachable call if the built-in cpu function is neither
16246 __builtin_cpu_is nor __builtin_cpu_supports.
16247 (rs6000_get_function_versions_dispatcher): Change the warning
16248 that an old GLIBC is used which does not export the capability
16249 bits to be an error.
16250 * doc/extend.texi (target_clones attribute): Document the
16251 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16252 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16253 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16254 the macros defined by GCC if the newer GLIBC is available.
16256 2017-07-12 Jeff Law <law@redhat.com>
16258 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16259 remaining includes slightly.
16260 * config/riscv/riscv-builtins.c: Include profile-count.h.
16262 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16265 * config/avr/avr.c (avr_set_current_function): In diagnostic
16266 messages: Quote keywords and (parts of) identifiers.
16267 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16270 2017-07-12 Carl Love <cel@us.ibm.com>
16272 * config/rs6000/rs6000-c.c: Add support for built-in functions
16273 vector bool char vec_revb (vector bool char);
16274 vector bool short vec_revb (vector short char);
16275 vector bool int vec_revb (vector bool int);
16276 vector bool long long vec_revb (vector bool long long);
16277 * doc/extend.texi: Update the built-in documentation file for the
16278 new built-in functions.
16280 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16282 * config/s390/s390.md: Remove movcc splitter.
16284 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16286 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16287 load/store on condition.
16289 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16292 * config/avr/avr.c (avr_encode_section_info)
16293 [progmem && !TREE_READONLY]: Error if progmem object needs
16296 2017-07-11 Michael Collison <michael.collison@arm.com>
16298 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16301 2017-07-11 Carl Love <cel@us.ibm.com>
16303 * config/rs6000/rs6000-c.c: Add support for builtins
16304 vector unsigned int vec_parity_lsbb (vector signed int);
16305 vector unsigned int vec_parity_lsbb (vector unsigned int);
16306 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16307 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16308 vector unsigned long long vec_parity_lsbb (vector signed long long);
16309 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16310 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16311 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16312 * doc/extend.texi: Update the built-in documentation file for the
16313 new built-in functions.
16315 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16317 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16318 (layout::m_primary_loc): New field.
16319 (layout::layout): Initialize new field. Move location filtering
16320 logic from here to...
16321 (layout::maybe_add_location_range): ...this new method. Add
16322 support for filtering to just the lines already specified by other
16324 (layout::will_show_line_p): New method.
16325 (gcc_rich_location::add_location_if_nearby): New method.
16326 (selftest::test_add_location_if_nearby): New test function.
16327 (selftest::diagnostic_show_locus_c_tests): Call it.
16328 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16331 2017-07-11 Tom de Vries <tom@codesourcery.com>
16333 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16334 (bb_first_real_insn): New function.
16335 (nvptx_single): Add extra initialization of broadcasted condition
16338 2017-07-11 Nathan Sidwell <nathan@acm.org>
16340 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16342 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16344 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16345 Binutils doc as TEXI will mess them up.
16346 * doc/invoke.texi (AVR Options): Same here.
16348 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16350 * config/sparc/sparc.opt (mfix-ut700): New option.
16351 (mfix-gr712rc): Likewise.
16352 (sparc_fix_b2bst): New variable.
16353 * doc/invoke.texi (SPARC options): Document them.
16354 (ARM options): Fix warnings.
16355 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16356 instructions to prevent sequences that can trigger the store-store
16357 errata for certain LEON3FT processors.
16358 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16359 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16360 * config/sparc/sparc.md (fix_b2bst): New attribute.
16361 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16363 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16366 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16368 (*rsqrtsf2_sse): Ditto.
16370 (div<mode>3): Macroize insn from divdf3 and divsf3
16371 using MODEF mode iterator.
16373 2017-07-10 Martin Sebor <msebor@redhat.com>
16375 PR tree-optimization/80397
16376 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16377 instead of testing for equality to INTEGER_TYPE.
16379 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16381 * config.gcc: Remove uclibc from arc target spec.
16383 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16385 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16387 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16390 * lto-wrapper.c (remove_option): New function.
16391 (merge_and_complain): Merge PIC/PIE options more realistically.
16393 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16395 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16399 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16400 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16401 * config.in: Regenerate.
16402 * configure: Regenerate.
16403 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16404 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16405 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16406 (TARGET_GASISR_PROLOGUES): ...target mask.
16407 * common/config/avr/avr-common.c
16408 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16409 Set -mgas-isr-prologues.
16410 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16411 INSERT_PASS_BEFORE for it.
16412 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16413 * config/avr/avr.c (avr_option_override)
16414 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16415 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16416 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16417 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16418 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16420 (make_avr_pass_pre_proep): New function.
16421 (emit_push_sfr) <treg>: Add argument to function and use it
16422 instead of TMP_REG.
16423 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16424 and set machine->gasisr.yes.
16425 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16426 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16427 __gcc_isr.n_pushed to .L__stack_usage.
16428 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16429 (avr_asm_final_postscan_insn): ...this new static function.
16430 * config/avr/avr.h (machine_function)
16431 <is_no_gccisr, use_L__stack_usage>: New fields.
16432 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16433 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16434 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16435 (gasisr, *gasisr): New expander and insn.
16436 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16437 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16438 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16440 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16442 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16445 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16447 Move jump-tables out of .text again.
16450 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16451 (ASM_OUTPUT_ADDR_VEC): New function.
16452 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16453 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16454 INSN_ADDRESSes as asm comment.
16455 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16456 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16457 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16458 * config/avr/avr.md (*tablejump): Adjust comment.
16459 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16460 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16462 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16463 (avr_output_addr_vec): New proto.
16464 (avr_log_t) <insn_addresses>: New field.
16466 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16469 * config/i386/i386.c (ix86_function_arg_advance): Set
16470 outgoing_args_on_stack to true if there are outgoing arguments
16472 (ix86_function_arg): Likewise.
16473 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16474 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16475 * config/i386/i386.h (machine_function): Add
16476 outgoing_args_on_stack.
16478 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16480 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16481 supporting pthreds.
16482 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16484 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16486 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16487 (REAL_H): Remove $(MACHMODE_H).
16488 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16490 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16491 $(MACHMODE_H) and double-int.h.
16492 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16494 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16497 2017-07-07 Andrew Pinski <apinski@cavium.com>
16499 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16500 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16502 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16504 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16505 Add warning if GCC was not configured to link against a GLIBC that
16506 exports the hardware capability bits.
16507 (make_resolver_func): Make resolver function private and not a
16508 COMDAT function. Create the name with clone_function_name instead
16509 of make_unique_name.
16512 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16513 correct operand in doing the split.
16515 2017-07-07 Carl Love <cel@us.ibm.com>
16517 * config/rs6000/rs6000-c: Add support for built-in function
16518 vector unsigned short vec_pack_to_short_fp32 (vector float,
16520 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16521 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16522 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16523 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16524 (convert_4f32_8i16): Add define_expand.
16525 * doc/extend.texi: Update the built-in documentation file for the
16526 new built-in function.
16528 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16530 * config/sparc/m8.md: New file.
16531 * config/sparc/sparc.md: Include m8.md.
16533 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16535 * config/sparc/sparc.opt: New option -mvis4b.
16536 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16537 (sparc_option_override): Handle VIS4B.
16538 (enum sparc_builtins): Define
16539 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16540 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16541 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16542 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16543 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16544 (check_constant_argument): New function.
16545 (sparc_vis_init_builtins): Define builtins
16546 __builtin_vis_dictunpack{8,16,32},
16547 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16548 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16549 __builtin_vis_fpcmpde{8,16,32}shl and
16550 __builtin_vis_fpcmpur{8,16,32}shl.
16551 (sparc_expand_builtin): Check that the constant operands to
16552 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16553 constant and in range.
16554 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16556 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16557 (SPARC_IMM5_P): Likewise.
16558 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16559 (enabled): Handle vis4b.
16560 (UNSPEC_DICTUNPACK): New unspec.
16561 (UNSPEC_FPCMPSHL): Likewise.
16562 (UNSPEC_FPUCMPSHL): Likewise.
16563 (UNSPEC_FPCMPDESHL): Likewise.
16564 (UNSPEC_FPCMPURSHL): Likewise.
16565 (cpu_feature): New CPU feature `vis4b'.
16566 (dictunpack{8,16,32}): New insns.
16567 (FPCSMODE): New mode iterator.
16568 (fpcscond): New code iterator.
16569 (fpcsucond): Likewise.
16570 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16571 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16572 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16573 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16574 * config/sparc/constraints.md: Define constraints `q' for unsigned
16575 2-bit integer constants and `t' for unsigned 5-bit integer
16577 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16579 (imm5_operand_dictunpack16): Likewise.
16580 (imm5_operand_dictunpack32): Likewise.
16581 (imm2_operand): Likewise.
16582 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16583 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16584 ditunpack* and fpcmp*shl builtins.
16586 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16588 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16589 * config.in: Add HAVE_AS_SPARC6 define.
16590 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16592 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16594 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16595 (CPP_CPU_SPEC): Handle m8.
16596 (ASM_CPU_SPEC): Likewise.
16597 * config/sparc/sparc-opts.h (enum processor_type): Add
16599 * config/sparc/sparc.c (m8_costs): New struct.
16600 (sparc_option_override): Handle TARGET_CPU_m8.
16601 (sparc32_initialize_trampoline): Likewise.
16602 (sparc64_initialize_trampoline): Likewise.
16603 (sparc_issue_rate): Likewise.
16604 (sparc_register_move_cost): Likewise.
16605 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16606 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16607 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16608 (CPP_CPU_SPEC): Handle M8.
16609 (ASM_CPU_SPEC): Likewise.
16610 (AS_M8_FLAG): Define.
16611 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16612 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16613 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16615 * configure: Regenerate.
16616 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16619 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16621 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16623 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16624 ("*movdi_insn_sp32"): Do not set v3pipe.
16625 ("*movsi_insn"): Likewise.
16626 ("*movdi_insn_sp64"): Likewise.
16627 ("*movsf_insn"): Likewise.
16628 ("*movdf_insn_sp32"): Likewise.
16629 ("*movdf_insn_sp64"): Likewise.
16630 ("*zero_extendsidi2_insn_sp64"): Likewise.
16631 ("*sign_extendsidi2_insn"): Likewise.
16632 ("*mov<VM32:mode>_insn"): Likewise.
16633 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16634 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16635 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16636 ("<vlop:code><VL:mode>3"): Likewise.
16637 ("*not_<vlop:code><VL:mode>3"): Likewise.
16638 ("*nand<VL:mode>_vis"): Likewise.
16639 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16640 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16641 ("one_cmpl<VL:mode>2"): Likewise.
16642 ("faligndata<VM64:mode>_vis"): Likewise.
16643 ("alignaddrsi_vis"): Likewise.
16644 ("alignaddrdi_vis"): Likweise.
16645 ("alignaddrlsi_vis"): Likewise.
16646 ("alignaddrldi_vis"): Likewise.
16647 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16648 ("bmaskdi_vis"): Likewise.
16649 ("bmasksi_vis"): Likewise.
16650 ("bshuffle<VM64:mode>_vis"): Likewise.
16651 ("cmask8<P:mode>_vis"): Likewise.
16652 ("cmask16<P:mode>_vis"): Likewise.
16653 ("cmask32<P:mode>_vis"): Likewise.
16654 ("pdistn<P:mode>_vis"): Likewise.
16655 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16657 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16659 * config/sparc/sparc.md ("subtype"): New insn attribute.
16660 ("*wrgsr_sp64"): Set insn subtype.
16661 ("*rdgsr_sp64"): Likewise.
16662 ("alignaddrsi_vis"): Likewise.
16663 ("alignaddrdi_vis"): Likewise.
16664 ("alignaddrlsi_vis"): Likewise.
16665 ("alignaddrldi_vis"): Likewise.
16666 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16667 ("fexpand_vis"): Likewise.
16668 ("fpmerge_vis"): Likewise.
16669 ("faligndata<VM64:mode>_vis"): Likewise.
16670 ("bshuffle<VM64:mode>_vis"): Likewise.
16671 ("cmask8<P:mode>_vis"): Likewise.
16672 ("cmask16<P:mode>_vis"): Likewise.
16673 ("cmask32<P:mode>_vis"): Likewise.
16674 ("fchksm16_vis"): Likewise.
16675 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16676 ("fmean16_vis"): Likewise.
16677 ("fp<plusminus_insn>64_vis"): Likewise.
16678 ("<plusminus_insn>v8qi3"): Likewise.
16679 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16680 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16681 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16682 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16683 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16684 ("*movqi_insn"): Likewise.
16685 ("*movhi_insn"): Likewise.
16686 ("*movsi_insn"): Likewise.
16687 ("movsi_pic_gotdata_op"): Likewise.
16688 ("*movdi_insn_sp32"): Likewise.
16689 ("*movdi_insn_sp64"): Likewise.
16690 ("movdi_pic_gotdata_op"): Likewise.
16691 ("*movsf_insn"): Likewise.
16692 ("*movdf_insn_sp32"): Likewise.
16693 ("*movdf_insn_sp64"): Likewise.
16694 ("*zero_extendhisi2_insn"): Likewise.
16695 ("*zero_extendqihi2_insn"): Likewise.
16696 ("*zero_extendqisi2_insn"): Likewise.
16697 ("*zero_extendqidi2_insn"): Likewise.
16698 ("*zero_extendhidi2_insn"): Likewise.
16699 ("*zero_extendsidi2_insn_sp64"): Likewise.
16700 ("ldfsr"): Likewise.
16701 ("prefetch_64"): Likewise.
16702 ("prefetch_32"): Likewise.
16703 ("tie_ld32"): Likewise.
16704 ("tie_ld64"): Likewise.
16705 ("*tldo_ldub_sp32"): Likewise.
16706 ("*tldo_ldub1_sp32"): Likewise.
16707 ("*tldo_ldub2_sp32"): Likewise.
16708 ("*tldo_ldub_sp64"): Likewise.
16709 ("*tldo_ldub1_sp64"): Likewise.
16710 ("*tldo_ldub2_sp64"): Likewise.
16711 ("*tldo_ldub3_sp64"): Likewise.
16712 ("*tldo_lduh_sp32"): Likewise.
16713 ("*tldo_lduh1_sp32"): Likewise.
16714 ("*tldo_lduh_sp64"): Likewise.
16715 ("*tldo_lduh1_sp64"): Likewise.
16716 ("*tldo_lduh2_sp64"): Likewise.
16717 ("*tldo_lduw_sp32"): Likewise.
16718 ("*tldo_lduw_sp64"): Likewise.
16719 ("*tldo_lduw1_sp64"): Likewise.
16720 ("*tldo_ldx_sp64"): Likewise.
16721 ("*mov<VM32:mode>_insn"): Likewise.
16722 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16723 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16725 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16727 * config/sparc/sparc.md ("type"): New insn type viscmp.
16728 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16730 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16731 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16732 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16733 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16735 ("n7_vis_logical_11cycle"): Likewise.
16736 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16737 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16738 * config/sparc/niagara.md ("niag_vis"): Likewise.
16739 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16740 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16742 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16744 * config/sparc/sparc.md: New instruction type `bmask'.
16745 (bmaskdi_vis): Use the `bmask' type.
16746 (bmasksi_vis): Likewise.
16747 * config/sparc/ultra3.md (us3_array): Likewise.
16748 * config/sparc/niagara7.md (n7_array): Likewise.
16749 * config/sparc/niagara4.md (n4_array): Likewise.
16750 * config/sparc/niagara2.md (niag2_vis): Likewise.
16751 (niag3_vis): Likewise.
16752 * config/sparc/niagara.md (niag_vis): Likewise.
16754 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16756 * ipa-comdats.c: Remove optimize check from gate.
16757 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16758 for functions not optimized.
16759 (ipa_fn_summary_read): Skip optimize check.
16760 (ipa_fn_summary_write): Likewise.
16761 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16763 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16765 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16766 (check_callers): Give up on caller which is not optimized.
16767 (inline_small_functions): Likewise.
16768 (ipa_inline): Do not give up when not optimizing.
16769 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16770 away unoptimizes cdtors.
16771 (whole_program_function_and_variable_visibility): Do
16772 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16773 * ipa.c (process_references): Do not check optimize.
16774 (symbol_table::remove_unreachable_nodes): Update optimize check.
16775 (set_writeonly_bit): Update optimize check.
16776 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16777 (pass_ipa_single_use::gate): Remove.
16779 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16781 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16782 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16783 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16784 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16785 permute_load, permute_store, adjust_extract, adjust_splat,
16786 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16787 replace_swap_with_copy, dump_swap_insn_table,
16788 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16789 recombine_lvx_pattern, recombine_stvx_pattern,
16790 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16791 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16792 to file rs6000-p8swap.c.
16793 * config/rs6000/rs6000-p8swap.c: New file.
16794 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16795 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16796 and rs6000*-*-* targets.
16798 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16800 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16802 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16804 * lto-wrapper.c (merge_and_complain): Do not merge
16805 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16806 fsigned_zeros, ftrapping_math, fwrapv.
16807 (append_compiler_options): Do not track these options.
16808 (append_linker_options): Likewie
16810 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16812 * cgraphunit.c (cgraph_node::finalize_function): When
16813 !flag_toplevel_reorde set no_reorder flag.
16814 (varpool_node::finalize_decl): Likewise.
16815 (symbol_table::compile): Drop no toplevel reorder path.
16817 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16819 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16820 edges; zero probability is not better than uninitialized.
16822 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16824 * asan.h (asan_sanitize_allocas_p): Declare.
16825 * asan.c (asan_sanitize_allocas_p): New function.
16826 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16827 (handle_builtin_alloca): Likewise.
16828 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16829 if !asan_sanitize_allocas_p.
16830 * params.def (asan-instrument-allocas): Add new option.
16831 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16832 * opts.c (common_handle_option): Disable allocas sanitization for
16835 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16837 * asan.c: Include gimple-fold.h.
16838 (get_last_alloca_addr): New function.
16839 (handle_builtin_stackrestore): Likewise.
16840 (handle_builtin_alloca): Likewise.
16841 (asan_emit_allocas_unpoison): Likewise.
16842 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16843 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16844 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16845 (instrument_builtin_call): Pass gimple iterator to
16846 get_mem_refs_of_builtin_call.
16847 (last_alloca_addr): New global.
16848 * asan.h (asan_emit_allocas_unpoison): Declare.
16849 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16850 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16851 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16852 if function calls alloca.
16853 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16854 * gimple-fold.h (replace_call_with_value): Declare.
16855 * internal-fn.c: Include asan.h.
16856 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16857 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16859 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16861 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16862 (C_SELFTEST_FLAGS): New.
16863 (CPP_SELFTEST_FLAGS): New.
16864 (SELFTEST_DEPS): New, from deps of s-selftest.
16865 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16866 (CPP_SELFTEST_DEPS): New.
16867 (selftest): Add dependency on s-selftest-c++.
16868 (s-selftest): Rename to...
16869 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16870 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16871 than SELFTEST_FLAGS.
16872 (selftest-gdb): Rename to...
16873 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16875 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16876 (selftest-valgrind): Rename to...
16877 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16879 (selftest-valgrind): Reintroduce as an alias for
16880 selftest-c-valgrind.
16881 (s-selftest-c++): New.
16882 (selftest-c++-gdb): New.
16883 (selftest-c++-valgrind): New.
16885 2017-07-06 Olivier Hainque <hainque@adacore.com>
16887 * gcc.c (process_command): When deciding if undefined variables
16888 should be ignored when processing specs, accept "gcc -v" as well.
16890 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16892 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16893 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16895 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16897 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16898 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16899 * config/arm/arm-tables.opt: Regenerate.
16900 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16902 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16904 2017-07-06 Carl Love <cel@us.ibm.com>
16906 * ChangeLog: Clean up from mid air collision
16908 2017-07-06 Carl Love <cel@us.ibm.com>
16910 * config/rs6000/rs6000-c.c: Add support for built-in functions
16911 vector signed int vec_subc (vector signed int, vector signed int);
16912 vector signed __int128 vec_subc (vector signed __int128,
16913 vector signed __int128);
16914 vector unsigned __int128 vec_subc (vector unsigned __int128,
16915 vector unsigned __int128);
16916 vector signed int vec_sube (vector signed int, vector signed int,
16917 vector signed int);
16918 vector unsigned int vec_sube (vector unsigned int,
16919 vector unsigned int,
16920 vector unsigned int);
16921 vector signed __int128 vec_sube (vector signed __int128,
16922 vector signed __int128,
16923 vector signed__int128);
16924 vector unsigned __int128 vec_sube (vector unsigned __int128,
16925 vector unsigned __int128,
16926 vector unsigned __int128);
16927 vector signed int vec_subec (vector signed int, vector signed int,
16928 vector signed int);
16929 vector unsigned int vec_subec (vector unsigned int,
16930 vector unsigned int,
16931 vector unsigned int);
16932 vector signed __int128 vec_subec (vector signed __int128,
16933 vector signed __int128,
16934 vector signed__int128);
16935 vector unsigned __int128 vec_subec (vector unsigned __int128,
16936 vector unsigned __int128,
16937 vector unsigned __int128);
16938 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16939 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16940 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16941 BU_ALTIVEC_OVERLOAD_X definitions.
16942 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16943 * doc/extend.texi: Update the built-in documentation file for the new
16944 built-in functions.
16946 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16949 * diagnostic-show-locus.c (layout::layout): Use start and finish
16950 spelling location for the start and finish of each range.
16951 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16952 Add unused aspect param.
16953 * input.c (expand_location_1): Add "aspect" param, and use it
16954 to access the correct part of the location.
16955 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16957 (expand_location_to_spelling_point): Likewise.
16958 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16959 param, and pass it to expand_location_1.
16961 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16963 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16964 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16965 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16966 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16967 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16968 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16969 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16970 _mm_maskz_getmant_ss): New intrinsics.
16971 (__builtin_ia32_getexpss128_mask): Changed to ...
16972 __builtin_ia32_getexpss128_round ... this.
16973 (__builtin_ia32_getexpsd128_mask): Changed to ...
16974 __builtin_ia32_getexpsd128_round ... this.
16975 * config/i386/i386-builtin-types.def
16976 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16977 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16978 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16979 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16980 __builtin_ia32_getmantss_mask_round): New builtins.
16981 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16982 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16983 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16984 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16985 * config/i386/sse.md
16986 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16987 avx512f_sgetexp<mode><mask_scalar_name>
16988 <round_saeonly_scalar_name> ... this.
16989 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16990 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16991 vgetexp<ssescalarmodesuffix>
16992 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16993 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16994 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16995 avx512f_vgetmant<mode><mask_scalar_name>
16996 <round_saeonly_scalar_name> ... this.
16997 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16998 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16999 vgetmant<ssescalarmodesuffix>
17000 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17001 %0<mask_scalar_operand4>, %1, %2
17002 <round_saeonly_scalar_mask_op4>, %3} ... this.
17003 * config/i386/subst.md (mask_scalar_operand4,
17004 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17005 round_saeonly_scalar_nimm_predicate): New subst attributes.
17007 2017-07-06 Julia Koval <julia.koval@intel.com>
17009 * config/i386/i386.c (ix86_erase_embedded_rounding):
17010 Remove code for old rounding pattern.
17012 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17014 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17016 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17018 * doc/sourcebuild.texi (Test Directives, Variants of
17019 dg-require-support): Add documentation for dg-require-stack-check.
17021 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17023 * config/i386/subst.md (mask_scalar, round_scalar,
17024 round_saeonly_scalar): New meta-templates.
17025 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17026 round_scalar_mask_operand3, round_scalar_mask_op3,
17027 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17028 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17029 round_saeonly_scalar_constraint,
17030 round_saeonly_scalar_prefix): New subst attribute.
17031 * config/i386/sse.md
17032 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17033 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17034 <round_scalar_name> ... this.
17035 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17036 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17037 <round_scalar_name> ... this.
17038 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17039 <sse>_vm<code><mode>3<mask_scalar_name>
17040 <round_saeonly_scalar_name> ... this.
17041 (v<plusminus_mnemonic><ssescalarmodesuffix>
17042 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17043 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17044 v<plusminus_mnemonic><ssescalarmodesuffix>
17045 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17046 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17047 (v<multdiv_mnemonic><ssescalarmodesuffix>
17048 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17049 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17050 v<multdiv_mnemonic><ssescalarmodesuffix>
17051 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17052 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17053 (v<maxmin_float><ssescalarmodesuffix>
17054 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17055 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17056 v<maxmin_float><ssescalarmodesuffix>
17057 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17058 %0<mask_scalar_operand3>, %1, %<iptr>2
17059 <round_saeonly_scalar_mask_op3>} ... this.
17061 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17063 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17064 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17066 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17067 Alan Hayward <alan.hayward@arm.com>
17068 David Sherwood <david.sherwood@arm.com>
17070 * combine.c (simplify_if_then_else): Remove "enum" before
17072 * compare-elim.c (can_eliminate_compare): Likewise.
17073 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17075 (aarch64_lookup_simd_builtin_type): Likewise.
17076 (aarch64_simd_builtin_type): Likewise.
17077 (aarch64_init_simd_builtin_types): Likewise.
17078 (aarch64_simd_expand_args): Likewise.
17079 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17081 (aarch64_reverse_mask): Likewise.
17082 (aarch64_simd_emit_reg_reg_move): Likewise.
17083 (aarch64_gen_adjusted_ldpstp): Likewise.
17084 (aarch64_ccmp_mode_to_code): Likewise.
17085 (aarch64_operands_ok_for_ldpstp): Likewise.
17086 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17087 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17089 (aarch64_min_divisions_for_recip_mul): Likewise.
17090 (aarch64_reassociation_width): Likewise.
17091 (aarch64_get_condition_code_1): Likewise.
17092 (aarch64_simd_emit_reg_reg_move): Likewise.
17093 (aarch64_simd_attr_length_rglist): Likewise.
17094 (aarch64_reverse_mask): Likewise.
17095 (aarch64_operands_ok_for_ldpstp): Likewise.
17096 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17097 (aarch64_gen_adjusted_ldpstp): Likewise.
17098 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17100 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17101 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17102 (arm_lookup_simd_builtin_type): Likewise.
17103 (arm_simd_builtin_type): Likewise.
17104 (arm_init_simd_builtin_types): Likewise.
17105 (arm_expand_builtin_args): Likewise.
17106 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17107 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17108 (ft32_setup_incoming_varargs): Likewise.
17109 (ft32_function_arg): Likewise.
17110 (ft32_function_arg_advance): Likewise.
17111 (ft32_pass_by_reference): Likewise.
17112 (ft32_arg_partial_bytes): Likewise.
17113 (ft32_valid_pointer_mode): Likewise.
17114 (ft32_addr_space_pointer_mode): Likewise.
17115 (ft32_addr_space_legitimate_address_p): Likewise.
17116 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17118 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17119 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17120 (iamcu_alignment): Likewise.
17121 (canonicalize_vector_int_perm): Likewise.
17122 (ix86_noce_conversion_profitable_p): Likewise.
17123 (ix86_mpx_bound_mode): Likewise.
17124 (ix86_operands_ok_for_move_multiple): Likewise.
17125 * config/microblaze/microblaze-protos.h
17126 (microblaze_expand_conditional_branch_reg): Likewise.
17127 * config/microblaze/microblaze.c
17128 (microblaze_expand_conditional_branch_reg): Likewise.
17129 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17131 (rs6000_reassociation_width): Likewise.
17132 (rs6000_invalid_binary_op): Likewise.
17133 (fusion_p9_p): Likewise.
17134 (emit_fusion_p9_load): Likewise.
17135 (emit_fusion_p9_store): Likewise.
17136 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17138 (riscv_hard_regno_mode_ok_p): Likewise.
17139 (riscv_address_insns): Likewise.
17140 (riscv_split_symbol): Likewise.
17141 (riscv_legitimize_move): Likewise.
17142 (riscv_function_value): Likewise.
17143 (riscv_hard_regno_nregs): Likewise.
17144 (riscv_expand_builtin): Likewise.
17145 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17146 (riscv_build_integer): Likewise.
17147 (riscv_split_integer): Likewise.
17148 (riscv_legitimate_constant_p): Likewise.
17149 (riscv_cannot_force_const_mem): Likewise.
17150 (riscv_regno_mode_ok_for_base_p): Likewise.
17151 (riscv_valid_base_register_p): Likewise.
17152 (riscv_valid_offset_p): Likewise.
17153 (riscv_valid_lo_sum_p): Likewise.
17154 (riscv_classify_address): Likewise.
17155 (riscv_legitimate_address_p): Likewise.
17156 (riscv_address_insns): Likewise.
17157 (riscv_load_store_insns): Likewise.
17158 (riscv_force_binary): Likewise.
17159 (riscv_split_symbol): Likewise.
17160 (riscv_force_address): Likewise.
17161 (riscv_legitimize_address): Likewise.
17162 (riscv_move_integer): Likewise.
17163 (riscv_legitimize_const_move): Likewise.
17164 (riscv_legitimize_move): Likewise.
17165 (riscv_address_cost): Likewise.
17166 (riscv_subword): Likewise.
17167 (riscv_output_move): Likewise.
17168 (riscv_canonicalize_int_order_test): Likewise.
17169 (riscv_emit_int_order_test): Likewise.
17170 (riscv_function_arg_boundary): Likewise.
17171 (riscv_pass_mode_in_fpr_p): Likewise.
17172 (riscv_pass_fpr_single): Likewise.
17173 (riscv_pass_fpr_pair): Likewise.
17174 (riscv_get_arg_info): Likewise.
17175 (riscv_function_arg): Likewise.
17176 (riscv_function_arg_advance): Likewise.
17177 (riscv_arg_partial_bytes): Likewise.
17178 (riscv_function_value): Likewise.
17179 (riscv_pass_by_reference): Likewise.
17180 (riscv_setup_incoming_varargs): Likewise.
17181 (riscv_print_operand): Likewise.
17182 (riscv_elf_select_rtx_section): Likewise.
17183 (riscv_save_restore_reg): Likewise.
17184 (riscv_for_each_saved_reg): Likewise.
17185 (riscv_register_move_cost): Likewise.
17186 (riscv_hard_regno_mode_ok_p): Likewise.
17187 (riscv_hard_regno_nregs): Likewise.
17188 (riscv_class_max_nregs): Likewise.
17189 (riscv_memory_move_cost): Likewise.
17190 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17191 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17192 (rl78_addr_space_address_mode): Likewise.
17193 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17195 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17196 (rs6000_reassociation_width): Likewise.
17197 (rs6000_invalid_binary_op): Likewise.
17198 (fusion_p9_p): Likewise.
17199 (emit_fusion_p9_load): Likewise.
17200 (emit_fusion_p9_store): Likewise.
17201 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17202 (ok_for_simple_move_operands): Likewise.
17203 (ok_for_simple_move_strict_operands): Likewise.
17204 (ok_for_simple_arith_logic_operands): Likewise.
17205 (visium_legitimize_reload_address): Likewise.
17206 (visium_select_cc_mode): Likewise.
17207 (output_cbranch): Likewise.
17208 (visium_split_double_move): Likewise.
17209 (visium_expand_copysign): Likewise.
17210 (visium_expand_int_cstore): Likewise.
17211 (visium_expand_fp_cstore): Likewise.
17212 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17213 (visium_function_arg): Likewise.
17214 (visium_function_arg_advance): Likewise.
17215 (visium_libcall_value): Likewise.
17216 (visium_setup_incoming_varargs): Likewise.
17217 (visium_legitimate_constant_p): Likewise.
17218 (visium_legitimate_address_p): Likewise.
17219 (visium_legitimize_address): Likewise.
17220 (visium_secondary_reload): Likewise.
17221 (visium_register_move_cost): Likewise.
17222 (visium_memory_move_cost): Likewise.
17223 (prepare_move_operands): Likewise.
17224 (ok_for_simple_move_operands): Likewise.
17225 (ok_for_simple_move_strict_operands): Likewise.
17226 (ok_for_simple_arith_logic_operands): Likewise.
17227 (visium_function_value_1): Likewise.
17228 (rtx_ok_for_offset_p): Likewise.
17229 (visium_legitimize_reload_address): Likewise.
17230 (visium_split_double_move): Likewise.
17231 (visium_expand_copysign): Likewise.
17232 (visium_expand_int_cstore): Likewise.
17233 (visium_expand_fp_cstore): Likewise.
17234 (visium_split_cstore): Likewise.
17235 (visium_select_cc_mode): Likewise.
17236 (visium_split_cbranch): Likewise.
17237 (output_cbranch): Likewise.
17238 (visium_print_operand_address): Likewise.
17239 * expmed.c (flip_storage_order): Likewise.
17240 * expmed.h (emit_cstore): Likewise.
17241 (flip_storage_order): Likewise.
17242 * genrecog.c (validate_pattern): Likewise.
17243 * hsa-gen.c (gen_hsa_addr): Likewise.
17244 * internal-fn.c (expand_arith_overflow): Likewise.
17245 * ira-color.c (allocno_copy_cost_saving): Likewise.
17246 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17247 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17248 (process_invariant_for_inheritance): Likewise.
17249 * lra-eliminations.c (move_plus_up): Likewise.
17250 * omp-low.c (lower_oacc_reductions): Likewise.
17251 * simplify-rtx.c (simplify_subreg): Likewise.
17252 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17253 (TARGET_CHKP_BOUND_MODE): Likewise..
17254 * targhooks.c (default_chkp_bound_mode): Likewise.
17255 (default_setup_incoming_vararg_bounds): Likewise.
17256 * targhooks.h (default_chkp_bound_mode): Likewise.
17257 (default_setup_incoming_vararg_bounds): Likewise.
17258 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17259 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17260 (have_whole_vector_shift): Likewise.
17261 * tree-vect-stmts.c (vectorizable_load): Likewise.
17262 * doc/tm.texi: Regenerate.
17264 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17266 Graceful degrade if Binutils PR21472 is not available.
17269 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17270 .rodata in flash test fails.
17271 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17272 * confgure: Regenerate.
17273 * config.in: Regenerate.
17274 * config/avr/avr.c (avr_asm_named_section)
17275 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17276 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17277 (avr_asm_init_sections): Same.
17279 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17281 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17282 (fma<VH:mode>4_intrinsic): Likewise.
17283 (*fmsub<VCVTF:mode>4): Likewise.
17284 (*fmsub<VH:mode>4_intrinsic): Likewise.
17286 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17289 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17290 Don't depend on "optimize > 0".
17291 (out_movhi_r_mr, out_movqi_mr_r): Same.
17292 (out_movhi_mr_r, out_movqi_r_mr): Same.
17293 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17294 io_address_operand on "optimize > 0".
17296 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17298 * tree-loop-distribution.c: Add general explanantion on the pass.
17299 (generate_loops_for_partition): Mark distributed loop.
17300 (pg_add_dependence_edges): New parameter. Handle alias data
17301 dependence specially and record it in the parameter if asked.
17302 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17303 (init_partition_graph_vertices, add_partition_graph_edge): New.
17304 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17305 (free_partition_graph_vdata, build_partition_graph): New.
17306 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17307 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17308 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17309 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17310 (version_for_distribution_p, finalize_partitions): New.
17311 (distribute_loop): Handle alias data dependence specially. Factor
17312 out loop fusion code as functions and call these functions.
17314 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17316 * tree-loop-distribution.c (classify_partition): New parameter and
17317 better handle reduction statement.
17318 (rdg_build_partitions): Revise comment.
17319 (distribute_loop): Compute statements in all partitions and pass it
17320 to classify_partition.
17322 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17324 * tree-loop-distribution.c (enum partition_type): New.
17325 (struct partition): New field type.
17326 (partition_merge_into): Add parameter. Update partition type.
17327 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17328 (build_rdg_partition_for_vertex): Compute partition type.
17329 (rdg_build_partitions): Dump partition type.
17330 (distribute_loop): Update calls to partition_merge_into.
17332 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17334 * tree-loop-distribution.c (struct ddr_hasher): New.
17335 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17337 (classify_partition): Call get_data_dependence.
17338 (pg_add_dependence_edges): Ditto.
17339 (distribute_loop): Release data dependence hash table.
17341 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17343 * tree-loop-distribution.c (ref_base_address): Delete.
17344 (similar_memory_accesses): Rename ...
17345 (share_memory_accesses): ... to this. Check if partitions access
17346 the same memory reference.
17347 (distribute_loop): Call share_memory_accesses.
17349 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17351 * tree-loop-distribution.c (struct partition): New field recording
17352 its data reference.
17353 (partition_alloc, partition_free): Init and release data refs.
17354 (partition_merge_into): Merge data refs.
17355 (build_rdg_partition_for_vertex): Collect data refs for partition.
17356 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17358 (distribute_loop): Remve data refs from vertice data of partition
17361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17363 * tree-loop-distribution.c (params.h): Include header file.
17364 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17365 (datarefs_vec): New global var.
17366 (create_rdg_vertices): Use datarefs_vec directly.
17367 (free_rdg): Don't free data references.
17368 (build_rdg): Update use. Don't free data references.
17369 (distribute_loop): Compute global variable for data references.
17370 Bail out if there are too many data references.
17372 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17374 * tree-loop-distribution.c (loop_nest): New global var.
17375 (build_rdg): Use loop directly, rather than loop nest.
17376 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17378 (distribute_loop): Compute global variable loop nest. Update use.
17380 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17382 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17383 (partition_merge_into): New parameter. Dump reason for fusion.
17384 (distribute_loop): Update use of partition_merge_into.
17386 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17388 * tree-loop-distribution.c (bb_top_order_index): New.
17389 (bb_top_order_index_size, bb_top_order_cmp): New.
17390 (stmts_from_loop): Use topological order.
17391 (pass_loop_distribution::execute): Compute and release topological
17392 order for basic blocks.
17394 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17396 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17399 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17401 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17402 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17403 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17404 * internal-fn.def (LOOP_DIST_ALIAS): New.
17405 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17406 (fold_loop_internal_call): ... this.
17407 (vect_loop_dist_alias_call): New function.
17408 (set_uid_loop_bbs): Call fold_loop_internal_call.
17409 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17412 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17415 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17416 Require dead FLAGS_REG at the beginning of a peephole.
17418 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17421 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17422 arguments in the call to __builtin_ia32_sbb_u32.
17423 (_subborrow_u64): Swap _X and _Y arguments in the call to
17424 __builtin_ia32_sbb_u64.
17426 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17429 * tree-vrp.c (compare_assert_loc): Turn into a function template
17430 with stable template parameter. Only test if a->e is NULL,
17431 !a->e == !b->e has been verified already. Use e == NULL or
17432 e != NULL instead of e or ! e tests. If stable is true, don't use
17433 iterative_hash_expr, on the other side allow a or b or both NULL
17434 and sort the NULLs last.
17435 (process_assert_insertions): Sort using compare_assert_loc<false>
17436 instead of compare_assert_loc, later sort using
17437 compare_assert_loc<true> before calling process_assert_insertions_for
17438 in a loop. Use break instead of continue once seen NULL pointer.
17440 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17442 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17443 Cortex-R7 and Cortex-R8 processors.
17445 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17447 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17448 uninitialized while src is not.
17450 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17452 * common/config/arm/arm-common.c: Adjust include path for
17454 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17455 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17456 (arm-cpu-data.h): Likewise.
17457 (arm-cpu-cdata.h): Likewise.
17458 * config/arm/arm-cpu.h: Delete.
17459 * config/arm/arm-cpu-cdata.h: Delete.
17460 * config/arm/arm-cpu-data.h: Delete.
17462 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17464 * config/arm/arm-cpus.in (cortex-a55): New.
17465 (cortex-a75): Likewise.
17466 (cortex-a75.cortex-a55): Likewise.
17467 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17469 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17470 * config/arm/arm-cpu-cdata.h: Regenerate.
17471 * config/arm/arm-cpu-data.h: Regenerate.
17472 * config/arm/arm-cpu.h: Regenerate.
17473 * config/arm/arm-tables.opt: Regenerate.
17474 * config/arm/arm-tune.md: Regenerate.
17476 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17478 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17480 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17482 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17485 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17487 PR tree-optimization/81292
17488 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17489 full_string_p, also call adjust_related_strinfos if the adjustment
17490 is simple, otherwise invalidate related strinfos.
17492 2017-07-04 Martin Liska <mliska@suse.cz>
17495 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17496 newly created variable as DECL_IGNORED_P.
17498 2017-07-04 Martin Liska <mliska@suse.cz>
17501 * ipa-inline.c (inline_small_functions):
17502 Use xstrdup_for_dump.
17504 2017-07-04 Tom de Vries <tom@codesourcery.com>
17506 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17508 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17511 * config/darwin.c (darwin_function_switched_text_sections):
17514 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17516 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17518 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17520 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17522 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17524 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17525 min_profitable_iters, and th as inclusive lower bounds.
17526 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17527 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17528 for min_profitable_iters and min_profitable_estimate.
17529 (vect_transform_loop): Treat th as an inclusive lower bound.
17530 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17532 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17535 * config/darwin.c (darwin_function_switched_text_sections):
17536 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17537 in two pieces, and suppress the use of buf.
17539 2017-07-03 Nathan Sidwell <nathan@acm.org>
17541 * hash-table.h (hash_table_mod1): Fix indentation.
17543 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17545 PR middle-end/81290
17546 * predict.c (force_edge_cold): Be more careful about propagation
17548 * profile-count.h (profile_probability::guessed,
17549 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17551 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17553 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17555 * doc/invoke.texi (rcpc architecture extension): Document it.
17557 2017-07-03 Richard Biener <rguenther@suse.de>
17559 PR tree-optimization/60510
17560 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17561 the scalar reduction PHI and use it.
17562 (vectorizable_reduction): Properly guard the single_defuse_cycle
17563 path for non-SLP reduction chains where we cannot use it.
17564 Rework reduc_def/index and vector type deduction. Rework
17565 vector operand gathering during reduction op code-gen.
17566 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17567 chains dissolve the chain and leave it to non-SLP reduction
17570 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17572 * tree-data-ref.h (dr_alignment): Declare.
17573 * tree-data-ref.c (dr_alignment): New function.
17574 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17575 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17577 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17579 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17581 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17582 and base_misalignment fields.
17583 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17584 * tree-data-ref.c: Include builtins.h.
17585 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17586 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17587 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17588 * tree-vect-data-refs.c: Include tree-cfg.h.
17589 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17590 fields instead of calculating an alignment here.
17591 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17592 innermost_loop_behavior fields.
17594 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17596 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17598 (DR_STEP_ALIGNMENT): New macro.
17599 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17600 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17601 (create_data_ref): Print it.
17602 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17603 to tell whether the step preserves vector (mis)alignment.
17604 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17605 Move the check for an integer step and generalise to all INTEGER_CST.
17606 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17607 Print the outer step alignment.
17609 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17611 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17612 with offset_alignment.
17613 (DR_ALIGNED_TO): Delete.
17614 (DR_OFFSET_ALIGNMENT): New macro.
17615 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17616 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17617 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17618 (create_data_ref): Likewise.
17619 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17620 (vect_analyze_data_refs): Likewise.
17621 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17622 creating dummy innermost behavior.
17624 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17626 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17627 with a "innermost_loop_behavior *" and refeence tree.
17628 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17629 (create_data_ref): Update call accordingly.
17630 * tree-predcom.c (find_looparound_phi): Likewise.
17632 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17634 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17635 fields with dr_wrt_vec_loop.
17636 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17637 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17638 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17639 (vect_dr_behavior): New function.
17640 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17641 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17642 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17643 track whether the step preserves the misalignment.
17644 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17645 Use vect_dr_behavior.
17646 (vect_setup_realignment): Update call accordingly.
17647 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17648 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17649 call to vect_create_addr_base_for_vector_ref.
17650 (vect_create_cond_for_align_checks): Likewise.
17651 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17652 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17653 (vect_recog_mask_conversion_pattern): Likewise.
17654 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17655 (new_stmt_vec_info): Remove redundant zeroing.
17657 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17659 * common/config/arm/arm-common.c (arm_be8_option): New function.
17660 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17661 (ISA_ARMv6): Add isa_bit_be8.
17662 * config/arm/arm.h (arm_be8_option): Add prototype.
17663 (BE8_SPEC_FUNCTION): New define.
17664 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17665 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17666 (mlittle-endian): Similarly.
17667 (mbe8, mbe32): New options.
17668 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17669 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17671 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17673 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17675 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17677 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17678 (cleanup_tree_cfg_bb): Use it.
17679 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17681 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17683 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17686 * loop-doloop.c (add_test): Update profile.
17688 2017-07-03 Martin Liska <mliska@suse.cz>
17691 * sanopt.c (rewrite_usage_of_param): New function.
17692 (sanitize_rewrite_addressable_params): Likewise.
17693 (pass_sanopt::execute): Call rewrite_usage_of_param.
17695 2017-07-03 Richard Biener <rguenther@suse.de>
17697 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17698 back to using VIEW_CONVERT_EXPR.
17700 2017-07-03 Martin Liska <mliska@suse.cz>
17703 * doc/extend.texi: Document when a resolver function is
17704 generated for target_clones.
17706 2017-07-03 Martin Liska <mliska@suse.cz>
17708 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17709 and shadow memory of auto variables which are subject of
17710 use-after-scope sanitization.
17711 (asan_expand_mark_ifn): Add do set only when is_poison.
17713 2016-07-03 Richard Biener <rguenther@suse.de>
17715 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17717 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17718 (vectorizable_reduction): Perform reduction PHI creation when
17719 visiting a reduction PHI and adjust and simplify code generation
17720 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17721 (vect_transform_loop): Visit reduction PHIs.
17722 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17723 defs into the SLP tree.
17724 (vect_build_slp_tree): Reduction defs terminate the recursion.
17725 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17727 (vect_get_vec_defs_for_stmt_copy): Export.
17728 (vect_get_vec_defs): Likewise.
17729 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17731 (vect_get_vec_defs_for_stmt_copy): Declare.
17732 (vect_get_vec_defs): Likewise.
17734 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17736 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17737 parameter with a "loop" parameter and use it instead of the
17738 loop containing DR_STMT. Don't check simple_iv when doing
17739 BB analysis. Describe the two analysis modes in the comment.
17741 2017-07-03 Tom de Vries <tom@codesourcery.com>
17743 PR tree-optimization/69468
17744 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17745 (find_same_succ_bb): Handle ignore_edge_flags.
17747 2017-07-03 Tom de Vries <tom@codesourcery.com>
17749 PR tree-optimization/81192
17750 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17752 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17754 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17756 2017-07-03 Tom de Vries <tom@codesourcery.com>
17758 PR tree-optimization/81192
17759 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17760 BB_SAME_SUCC (bb) == NULL.
17762 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17764 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17767 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17769 * dumpfile.c: Include profile-count.h
17770 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17772 (insert_cond_bb): Update profile.
17773 * tree-cfg.h (insert_cond_bb): Update prototype.
17774 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17775 * tree-dump.c: Do not include tree-cfg.
17777 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17779 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17781 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17783 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17786 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17788 * tree-complex.c (expand_complex_div_wide): update profile.
17790 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17791 Alan Hayward <alan.hayward@arm.com>
17792 David Sherwood <david.sherwood@arm.com>
17794 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17795 (CORETYPES_H): New define.
17796 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17797 (insn-modes-inline.h, s-modes-inline-h): New rules.
17798 (generated_files): Add insn-modes-inline.h.
17799 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17800 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17801 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17802 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17803 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17804 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17805 (build/gencodes.o, build/genconditions.o): Likewise.
17806 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17807 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17808 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17809 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17810 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17811 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17812 * coretypes.h: Include everything up to real.h for generators.
17813 Include insn-modes.h first. Include wide-int-print.h after
17814 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17815 * machmode.h: Don't include insn-modes.h here.
17816 * function-tests.c: Remove includes of signop.h, machmode.h,
17817 double-int.h and wide-int.h.
17819 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17821 * optc-save-gen.awk: Likewise.
17822 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17823 * godump.c: Remove include of wide-int-print.h.
17824 * pretty-print.h: Likewise.
17825 * wide-int-print.cc: Likewise.
17826 * wide-int.cc: Likewise.
17827 * hash-map-tests.c: Remove include of signop.h.
17828 * hash-set-tests.c: Likewise.
17829 * rtl-tests.c: Likewise.
17830 * mkconfig.sh: Remove include of machmode.h.
17831 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17833 (emit_insn_modes_inline_h): ...this new function. Emit the code
17834 into an insn-modes-inline.h header file, adding appropriate
17835 include guards and end comments.
17836 (emit_insn_modes_c_header): Remove include of machmode.h.
17837 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17839 (main): Handle -i flag and call emit_insn_modes_inline_h when
17842 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17844 * tree-ssa-strlen.c (strinfo): Rename the length field to
17845 nonzero_chars. Add a full_string_p field.
17846 (compare_nonzero_chars, zero_length_string_p): New functions.
17847 (get_addr_stridx): Add an offset_out parameter.
17848 Use compare_nonzero_chars.
17849 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17850 (new_strinfo): Update after above changes to strinfo.
17851 (set_endptr_and_length): Set full_string_p.
17852 (get_string_length): Update after above changes to strinfo.
17853 (unshare_strinfo): Update call to new_strinfo.
17854 (maybe_invalidate): Likewise.
17855 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17856 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17857 as a uhwi instead of an shwi. Update after above changes to
17858 strinfo and new_strinfo.
17859 (zero_length_string): Assert that chainsi contains full strings.
17860 Use zero_length_string_p. Update call to new_strinfo.
17861 (adjust_related_strinfos): Update after above changes to strinfo.
17862 Copy full_string_p from origsi.
17863 (adjust_last_stmt): Use zero_length_string_p.
17864 (handle_builtin_strlen): Update after above changes to strinfo and
17865 new_strinfo. Install the lhs as the string length if the previous
17866 entry didn't describe a full string.
17867 (handle_builtin_strchr): Update after above changes to strinfo
17869 (handle_builtin_strcpy): Likewise.
17870 (handle_builtin_strcat): Likewise.
17871 (handle_builtin_malloc): Likewise.
17872 (handle_pointer_plus): Likewise.
17873 (handle_builtin_memcpy): Likewise. Track nonzero characters
17874 that aren't necessarily followed by a nul terminator.
17875 (handle_char_store): Likewise.
17877 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17879 PR tree-optimization/80769
17880 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17881 for malloc and calloc. Document the new invariant that all related
17882 strinfos have delayed lengths or none do.
17883 (verify_related_strinfos): Move earlier in file.
17884 (set_endptr_and_length): New function, split out from...
17885 (get_string_length): ...here. Also set the lengths of related
17887 (zero_length_string): Assert that chainsi has known (rather than
17889 (adjust_related_strinfos): Likewise.
17891 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17893 PR tree-optimization/81136
17894 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17895 assert that two references with the same misalignment have the same
17896 compile-time misalignment if those compile-time misalignments
17899 2017-07-01 Andi Kleen <ak@linux.intel.com>
17901 * print-tree.c (print_node): Print all attributes.
17903 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17905 * cfg.c (scale_bbs_frequencies): New function.
17906 * cfg.h (scale_bbs_frequencies): Declare it.
17907 * cfgloopanal.c (single_likely_exit): Cleanup.
17908 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17910 (scale_loop_profile): Likewise.
17911 (loop_version): Likewise.
17912 (create_empty_loop_on_edge): Update.
17913 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17914 scale_loop_frequencies, scale_loop_profile, loopify,
17915 loop_version): Update prototypes.
17916 * modulo-sched.c (sms_schedule): Update.
17917 * predict.c (unlikely_executed_edge_p): Also check probability.
17918 (probably_never_executed_edge_p): Fix typo.
17919 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17920 * tree-parloops.c (gen_parallel_loop): Update.
17921 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17922 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17923 * tree-ssa-loop-split.c (split_loop): Update.
17924 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17925 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17926 (vect_loop_versioning): Update.
17927 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17929 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17931 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17933 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17935 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17936 to keep profile consistent.
17938 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17940 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17941 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17942 * profile-count.h (max_safe_multiplier): Make unsigned.
17943 (profile_count::guessed_zero): New.
17945 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17947 * bb-reorder.c (fix_up_crossing_landing_pad,
17948 fix_crossing_conditional_branches): Use make_single_succ_edge
17949 to keep profile consistent.
17951 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17953 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17956 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17959 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17960 the right scopes, make sure cond_jump isn't preserved between multiple
17961 iterations. Search for fallthru edge whenever there are 3+ edges and
17962 use find_fallthru_edge for it.
17964 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17966 Patch by Alexander Monakov <amonakov@ispras.ru>
17967 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17968 probabilities consistently.
17970 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17972 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17973 emit_cmp_and_jump_insns.
17975 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17978 * tree-inline.c (expand_call_inline): Combine profile statuses.
17980 2017-06-30 Andrew Pinski <apinski@cavium.com>
17982 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17983 fold_stmt returned true.
17985 2017-06-30 Nathan Sidwell <nathan@acm.org>
17987 * ggc.h (empty_string): Delete.
17988 * cfgexpand.c (expand_asm_stmt): Use plain "".
17989 * optabs.c (expand_asm_memory_barrier): Likewise.
17990 * stringpool.c (empty_string): Delete.
17991 (digit_vector, digit_string): Delete.
17992 (ggc_alloc_string): Use plain "", don't optimize single digit
17993 strings. Use ggc_alloc_atomic.
17995 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17997 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17998 comparison set and one other set, use the cost of the non-comparison
18001 2017-06-30 Nathan Sidwell <nathan@acm.org>
18003 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18006 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18008 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18009 loops. Remove now unneeded calls to gimple_switch_set_label() that
18010 just set removed labels to NULL_TREE.
18012 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18014 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18015 (set_range_info): ...here. Only call set_range_info_raw if domain
18017 (set_nonzero_bits): Call set_range_info_raw.
18018 * tree-ssanames.h (set_range_info_raw): New.
18020 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18023 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18024 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18025 of nonimmediate_operand and <store_mask_constraint> instead of m
18026 for the input operand. For V8FI iterator, always split if input
18027 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18028 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18029 <store_mask_predicate> instead of register_operand and
18030 <store_mask_constraint> instead of v for the input operand. Make
18031 sure both operands aren't MEMs for if not <mask_applied>.
18033 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18035 * lto-wrapper.c (copy_file) Close both file descriptors before
18038 2017-06-30 Martin Liska <mliska@suse.cz>
18041 * multiple_target.c (create_dispatcher_calls): Make ifunc
18042 also for function that don't have calls or are not referenced.
18044 2017-06-30 Richard Biener <rguenther@suse.de>
18046 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18047 analyze the first scalar stmt. Move vector type computation
18048 for the BB case here from ...
18049 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18050 live operation processing in the SLP case properly.
18052 2017-06-30 Richard Biener <rguenther@suse.de>
18054 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18056 2017-06-30 Martin Liska <mliska@suse.cz>
18059 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18060 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18062 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18064 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18067 2017-06-30 Richard Biener <rguenther@suse.de>
18069 PR tree-optimization/81249
18070 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18071 condition reduction result to original scalar type.
18073 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18075 * profile-count.h (enum profile_quality): Fix typos and whitespace
18078 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18080 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18081 type for branch probabilities.
18083 2017-06-29 Julian Brown <julian@codesourcery.com>
18084 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18086 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18087 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18088 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18089 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18091 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18093 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18094 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18095 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18096 CC usage from generic code to here.
18097 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18098 CC usage into the target macros.
18100 2017-06-29 Maya Rashish <coypu@sdf.org>
18102 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18105 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18107 * arm/arm-builtins.c: Include profile-count.h
18108 * except.c (sjlj_emit_function_enter): Use
18109 profile_probability::unlikely.
18111 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18113 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18114 and tocrel_offset be pointer args rather than implicitly using
18116 (legitimate_constant_pool_address_p, rs6000_emit_move,
18117 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18118 tocrel_offset and use in toc_relative_expr_p call.
18119 (print_operand, print_operand_address): Use static tocrel_base_oac
18120 and tocrel_offset_oac.
18121 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18124 2017-06-29 Maya Rashish <coypu@sdf.org>
18126 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18128 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18130 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18131 objects, take into account only the alignment of 'op0' and 'mode1' if
18134 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18136 * ccmp.c (ccmp_tree_comparison_p): New function.
18137 (ccmp_candidate_p): Update to use above function.
18138 (get_compare_parts): New function.
18139 (expand_ccmp_next): Update to use new functions.
18140 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18142 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18143 take mode as argument.
18144 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18145 * expr.c (expand_expr_real_1): Pass mode as argument.
18147 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18149 * combine.c (combine_instructions): Print insns to dump_file, together
18152 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18154 * asan.c (asan_emit_stack_protection): Update.
18155 (create_cond_insert_point): Update.
18156 * auto-profile.c (afdo_propagate_circuit): Update.
18157 * basic-block.h (struct edge_def): Turn probability to
18158 profile_probability.
18159 (EDGE_FREQUENCY): Update.
18160 * bb-reorder.c (find_traces_1_round): Update.
18161 (better_edge_p): Update.
18162 (sanitize_hot_paths): Update.
18163 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18164 (make_single_succ_edge): Update.
18165 (check_bb_profile): Update.
18166 (dump_edge_info): Update.
18167 (update_bb_profile_for_threading): Update.
18168 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18169 probabilitycount to 0.
18170 * cfgbuild.c (compute_outgoing_frequencies): Update.
18171 * cfgcleanup.c (try_forward_edges): Update.
18172 (outgoing_edges_match): Update.
18173 (try_crossjump_to_edge): Update.
18174 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18175 (expand_gimple_tailcall): Update.
18176 (construct_init_block): Use make_single_succ_edge.
18177 (construct_exit_block): Use make_single_succ_edge.
18178 * cfghooks.c (verify_flow_info): Update.
18179 (redirect_edge_succ_nodup): Update.
18180 (split_edge): Update.
18181 (account_profile_record): Update.
18182 * cfgloopanal.c (single_likely_exit): Update.
18183 * cfgloopmanip.c (scale_loop_profile): Update.
18184 (set_zero_probability): Remove.
18185 (duplicate_loop_to_header_edge): Update.
18186 * cfgloopmanip.h (loop_version): Update prototype.
18187 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18188 (force_nonfallthru_and_redirect): Update.
18189 (update_br_prob_note): Update.
18190 (rtl_verify_edges): Update.
18191 (purge_dead_edges): Update.
18192 (rtl_lv_add_condition_to_bb): Update.
18193 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18194 * cgraphunit.c (init_lowered_empty_function): Update.
18195 (cgraph_node::expand_thunk): Update.
18196 * cilk-common.c: Include profile-count.h
18197 * dojump.c (inv): Remove.
18198 (jumpifnot): Update.
18199 (jumpifnot_1): Update.
18200 (do_jump_1): Update.
18202 (do_jump_by_parts_greater_rtx): Update.
18203 (do_compare_rtx_and_jump): Update.
18204 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18205 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18206 * dwarf2cfi.c: Include profile-count.h
18207 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18208 (sjlj_emit_dispatch_table): Likewise.
18209 * explow.c: Include profile-count.h
18210 * expmed.c (emit_store_flag_force): Update.
18211 (do_cmp_and_jump): Update.
18212 * expr.c (compare_by_pieces_d::generate): Update.
18213 (compare_by_pieces_d::finish_mode): Update.
18214 (emit_block_move_via_loop): Update.
18215 (store_expr_with_bounds): Update.
18216 (store_constructor): Update.
18217 (expand_expr_real_2): Update.
18218 (expand_expr_real_1): Update.
18219 * expr.h (try_casesi, try_tablejump): Update prototypes.
18220 * gimple-pretty-print.c (dump_probability): Update.
18221 (dump_profile): New.
18222 (dump_gimple_label): Update.
18223 (dump_gimple_bb_header): Update.
18224 * graph.c (draw_cfg_node_succ_edges): Update.
18225 * hsa-gen.c (convert_switch_statements): Update.
18226 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18227 (find_if_case_1): Update.
18228 (find_if_case_2): Update.
18229 * internal-fn.c (expand_arith_overflow_result_store): Update.
18230 (expand_addsub_overflow): Update.
18231 (expand_neg_overflow): Update.
18232 (expand_mul_overflow): Update.
18233 (expand_vector_ubsan_overflow): Update.
18234 * ipa-cp.c (good_cloning_opportunity_p): Update.
18235 * ipa-split.c (split_function): Use make_single_succ_edge.
18236 * ipa-utils.c (ipa_merge_profiles): Update.
18237 * loop-doloop.c (add_test): Update.
18238 (doloop_modify): Update.
18239 * loop-unroll.c (compare_and_jump_seq): Update.
18240 (unroll_loop_runtime_iterations): Update.
18241 * lra-constraints.c (lra_inheritance): Update.
18242 * lto-streamer-in.c (input_cfg): Update.
18243 * lto-streamer-out.c (output_cfg): Update.
18244 * mcf.c (adjust_cfg_counts): Update.
18245 * modulo-sched.c (sms_schedule): Update.
18246 * omp-expand.c (expand_omp_for_init_counts): Update.
18247 (extract_omp_for_update_vars): Update.
18248 (expand_omp_ordered_sink): Update.
18249 (expand_omp_for_ordered_loops): Update.
18250 (expand_omp_for_generic): Update.
18251 (expand_omp_for_static_nochunk): Update.
18252 (expand_omp_for_static_chunk): Update.
18253 (expand_cilk_for): Update.
18254 (expand_omp_simd): Update.
18255 (expand_omp_taskloop_for_outer): Update.
18256 (expand_omp_taskloop_for_inner): Update.
18257 * omp-simd-clone.c (simd_clone_adjust): Update.
18258 * optabs.c (expand_doubleword_shift): Update.
18259 (expand_abs): Update.
18260 (emit_cmp_and_jump_insn_1): Update.
18261 (expand_compare_and_swap_loop): Update.
18262 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18263 * predict.c (predictable_edge_p): Update.
18264 (edge_probability_reliable_p): Update.
18265 (set_even_probabilities): Update.
18266 (combine_predictions_for_insn): Update.
18267 (combine_predictions_for_bb): Update.
18268 (propagate_freq): Update.
18269 (estimate_bb_frequencies): Update.
18270 (force_edge_cold): Update.
18271 * profile-count.c (profile_count::dump): Add missing space into dump.
18272 (profile_count::debug): Add newline.
18273 (profile_count::differs_from_p): Explicitly convert to unsigned.
18274 (profile_count::stream_in): Update.
18275 (profile_probability::dump): New member function.
18276 (profile_probability::debug): New member function.
18277 (profile_probability::differs_from_p): New member function.
18278 (profile_probability::differs_lot_from_p): New member function.
18279 (profile_probability::stream_in): New member function.
18280 (profile_probability::stream_out): New member function.
18281 * profile-count.h (profile_count_quality): Rename to ...
18282 (profile_quality): ... this one.
18283 (profile_probability): New.
18284 (profile_count): Update.
18285 * profile.c (compute_branch_probabilities): Update.
18286 * recog.c (peep2_attempt): Update.
18287 * sched-ebb.c (schedule_ebbs): Update.
18288 * sched-rgn.c (find_single_block_region): Update.
18289 (compute_dom_prob_ps): Update.
18290 (schedule_region): Update.
18291 * sel-sched-ir.c (compute_succs_info): Update.
18292 * stmt.c (struct case_node): Update.
18293 (do_jump_if_equal): Update.
18294 (get_outgoing_edge_probs): Update.
18295 (conditional_probability): Update.
18296 (emit_case_dispatch_table): Update.
18297 (expand_case): Update.
18298 (expand_sjlj_dispatch_table): Update.
18299 (emit_case_nodes): Update.
18300 * targhooks.c: Update.
18301 * tracer.c (better_p): Update.
18302 (find_best_successor): Update.
18303 * trans-mem.c (expand_transaction): Update.
18304 * tree-call-cdce.c: Update.
18305 * tree-cfg.c (gimple_split_edge): Upate.
18306 (move_sese_region_to_fn): Upate.
18307 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18308 * tree-eh.c (lower_resx): Upate.
18309 (cleanup_empty_eh_move_lp): Upate.
18310 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18311 * tree-inline.c (copy_edges_for_bb): Update.
18312 (copy_cfg_body): Update.
18313 * tree-parloops.c (gen_parallel_loop): Update.
18314 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18315 (gimple_gen_time_profiler): Update.
18316 * tree-ssa-dce.c (remove_dead_stmt): Update.
18317 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18318 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18319 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18320 (unloop_loops): Update.
18321 (try_peel_loop): Update.
18322 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18323 * tree-ssa-loop-split.c (connect_loops): Update.
18324 (split_loop): Update.
18325 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18326 (hoist_guard): Update.
18327 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18328 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18329 (value_replacement): Update.
18330 * tree-ssa-reassoc.c (branch_fixup): Update.
18331 * tree-ssa-tail-merge.c (replace_block_by): Update.
18332 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18333 (create_edge_and_update_destination_phis): Update.
18334 (compute_path_counts): Update.
18335 (recompute_probabilities): Update.
18336 (update_joiner_offpath_counts): Update.
18337 (freqs_to_counts_path): Update.
18338 (duplicate_thread_path): Update.
18339 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18340 (struct switch_conv_info): Update.
18341 (gen_inbound_check): Update.
18342 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18343 (vect_do_peeling): Update.
18344 (vect_loop_versioning): Update.
18345 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18346 (optimize_mask_stores): Update.
18347 * ubsan.c (ubsan_expand_null_ifn): Update.
18348 * value-prof.c (gimple_divmod_fixed_value): Update.
18349 (gimple_divmod_fixed_value_transform): Update.
18350 (gimple_mod_pow2): Update.
18351 (gimple_mod_pow2_value_transform): Update.
18352 (gimple_mod_subtract): Update.
18353 (gimple_mod_subtract_transform): Update.
18354 (gimple_ic): Update.
18355 (gimple_stringop_fixed_value): Update.
18356 (gimple_stringops_transform): Update.
18357 * value-prof.h: Update.
18359 2017-06-29 Carl Love <cel@us.ibm.com>
18361 * config/rs6000/rs6000-c.c: Add support for built-in functions
18362 vector signed int vec_signed (vector float);
18363 vector signed long long vec_signed (vector double);
18364 vector signed int vec_signed2 (vector double, vector double);
18365 vector signed int vec_signede (vector double);
18366 vector signed int vec_signedo (vector double);
18367 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18368 instruction generator.
18369 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18370 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18371 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18373 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18374 vunsignede_v2df): Add define_expands.
18375 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18376 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18377 VEC_UNSIGNEDO): Add definitions.
18378 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18379 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18380 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18381 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18382 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18383 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18384 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18385 vec_unsignede, vec_unsignedo): Add builtin defines.
18386 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18388 * doc/extend.texi: Update the built-in documentation file for the
18389 new built-in functions.
18391 2017-06-29 Richard Biener <rguenther@suse.de>
18393 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18394 reduction chains to LOOP_VINFO_REDUCTIONS.
18395 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18396 SLP reductions after processing reduction chains.
18398 2017-06-29 Nathan Sidwell <nathan@acm.org>
18400 * builtins.c (fold_builtin_FUNCTION): Use
18401 lang_hooks.decl_printable_name.
18403 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18405 PR middle-end/81194
18406 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18407 with only one label.
18408 * stmt.c (expand_case): Assert NCASES is greater than one.
18410 2017-06-29 Richard Biener <rguenther@suse.de>
18412 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18414 (group_case_labels): Likewise.
18415 (find_taken_edge): Push sanity checking on val to workers...
18416 (find_taken_edge_cond_expr): ... here
18417 (find_taken_edge_switch_expr): ... and here, handle cases
18418 with just a default label.
18419 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18420 (group_case_labels): Likewise.
18421 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18422 group_case_labels does anything cleanup the CFG again.
18424 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18426 PR tree-optimization/81196
18427 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18428 exit condition comparing two IVs.
18430 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18432 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18433 profile to the dummy entry at the end of the list of architectures.
18434 * config/arm/arm-cpu-cdata.h: Regenerated.
18436 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18437 Michael Collison <michael.collison@arm.com>
18440 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18442 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18443 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18444 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18445 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18446 of shift when the shift amount is masked with constant equal to
18447 the size of the mode.
18448 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18451 2017-06-29 Martin Liska <mliska@suse.cz>
18453 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18455 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18458 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18461 (edge_growth_cache_entry::edge_growth_cache_entry): New
18463 (reset_edge_growth_cache): Update to use constructor.
18465 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18467 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18468 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18469 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18471 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18473 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18474 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18476 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18478 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18479 (*-linux-uclibc*): Add t-uclibc tmake_file.
18480 * config/t-musl: New.
18481 * config/t-uclibc: New.
18483 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18485 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18487 (gen_comm_data): Emit architectural setting of arch_prof.
18488 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18490 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18491 (armv8-m.base, armv8-m.main): Likewise.
18492 * arm-protos.h (arm_build_target): Add profile field.
18493 (arch_option): Likewise.
18494 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18496 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18497 arm_active_target.profile.
18499 2017-06-28 Richard Biener <rguenther@suse.de>
18501 PR middle-end/81227
18502 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18503 TYPE_OVERFLOW_WRAPS.
18504 * match.pd (negate_expr_p): Likewise.
18505 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18506 fold_build2, not fold_binary.
18508 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18510 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18511 Convert memory address to Pmode.
18512 (aarch64_print_operand): Assert MEM operands are always Pmode.
18514 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18517 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18518 Remove redundant if.
18519 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18520 * config/arm/aarch-common-protos.h
18521 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18522 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18524 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18527 * multiple_target.c (create_dispatcher_calls): Set the default
18528 clone to be static, not public.
18530 2017-06-28 Richard Biener <rguenther@suse.de>
18532 * tree-vect-loop.c (vectorizable_reduction): Move special
18533 cond reduction IV var creation ...
18534 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18535 parameter. Use STMT_VINFO_VECTYPE.
18536 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18539 2017-06-28 Martin Liska <mliska@suse.cz>
18542 * ipa-visibility.c (non_local_p): Handle visibility.
18544 2017-06-28 Martin Liska <mliska@suse.cz>
18547 * common.opt: Add IntegerRange to various options.
18548 * opt-functions.awk (integer_range_info): New function.
18549 * optc-gen.awk: Add integer_range_info to cl_options struct.
18550 * opts-common.c (decode_cmdline_option): Handle
18551 CL_ERR_INT_RANGE_ARG.
18552 (cmdline_handle_error): Likewise.
18553 * opts.c (print_filtered_help): Show valid interval in
18554 when --help is provided.
18555 * opts.h (struct cl_option): Add range_min and range_max fields.
18556 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18558 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18560 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18561 (x * C EQ/NE y * C): New transformation.
18563 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18565 * genmultilib (combination_space): Accept '+' in option names.
18567 2017-06-28 Martin Liska <mliska@suse.cz>
18570 * asan.c (instrument_derefs): Bail out inner references
18571 that are hard register variables.
18573 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18576 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18577 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18579 2017-06-28 Richard Biener <rguenther@suse.de>
18581 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18582 (vect_get_slp_defs): Adjust.
18583 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18585 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18587 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18588 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18589 (vectorizable_reduction): Adjust.
18590 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18592 (vect_get_slp_defs): Likewise.
18593 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18594 (vectorizable_bswap): Adjust.
18595 (vectorizable_call): Likewise.
18596 (vectorizable_conversion): Likewise.
18597 (vectorizable_assignment): Likewise.
18598 (vectorizable_shift): Likewise.
18599 (vectorizable_operation): Likewise.
18600 (vectorizable_store): Likewise.
18601 (vectorizable_condition): Likewise.
18602 (vectorizable_comparison): Likewise.
18604 2017-06-28 Michael Collison <michael.collison@arm.com>
18607 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18609 (arm_gen_movmemqi): Removed unused variable 'i'.
18610 Convert 'for' loop into 'while' loop.
18611 (arm_expand_prologue): Remove last unnecessary set of insn.
18612 (thumb_pop): Remove unused variable 'pushed_words'.
18613 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18615 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18617 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18618 * config/s390/s390-protos.h: Add prototype of
18619 s390_rel_address_ok_p.
18620 * config/s390/s390.c (s390_got_symbol): New function.
18621 (s390_rel_address_ok_p): New function.
18622 (legitimize_pic_address): Use s390_rel_address_ok_p.
18623 (s390_load_got): Use s390_got_symbol.
18624 (s390_option_override): Issue error if
18625 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18626 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18628 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18630 2017-06-27 Andrew Pinski <apinski@cavium.com>
18632 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18633 (X * copysign (1.0, X)): New pattern.
18634 (X * copysign (1.0, -X)): New pattern.
18635 (copysign (-1.0, CST)): New pattern.
18637 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18639 * genmultilib (combination_space): Remove variable.
18640 Validate reuse rules against regular expression for any sequence
18641 of multilib options in any order.
18643 2017-06-27 Michael Collison <michael.collison@arm.com>
18645 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18646 call aarch64_split_simd_combine.
18647 * (aarch64_combine_internal<mode>): Delete pattern.
18648 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18649 Allow register and subreg operands.
18651 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18653 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18654 specific need, just fallback on defaults.
18655 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18657 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18658 Olivier Hainque <hainque@adacore.com>
18660 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18662 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18663 targets. Pick a default if no particular attempt applied.
18664 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18667 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18669 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18670 (x86_64-wrs-vxworks7): Likewise.
18672 2017-06-27 Marek Polacek <polacek@redhat.com>
18675 * ubsan.c (instrument_null): Check get_base_address's result for null.
18677 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18679 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18681 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18683 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18684 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18685 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18686 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18687 New function types.
18688 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18689 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18690 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18691 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18692 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18693 BUILT_IN_FEUPDATEENV): New builtins.
18694 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18695 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18696 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18697 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18699 (builtin_structptr_types): Adjust size.
18700 * tree.c (builtin_structptr_types): Add four entries.
18702 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18703 Olivier Hainque <hainque@adacore.com>
18705 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18706 (TLS_SYM): New local macro, forcing reference to __tls__ on
18707 link command lines for VxWorks 7 RTPs, triggering initialization
18709 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18710 OS features TLS support, true for RTPs on VxWorks 7.
18711 * config/vxworks.c (vxworks_override_options): Setup emutls
18714 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18716 * predict.c (test_prediction_value_range): Use -1U instead of -1
18717 to avoid narrowing conversion warning.
18718 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18719 to avoid narrowing conversion warning.
18720 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18722 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18724 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18726 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18727 64bit configurations.
18728 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18729 (SIZE_TYPE): Likewise.
18730 * config/vxworks.c (vxworks_emutls_var_fields): Use
18731 long_unsigned_type_node instead of unsigned_type_node as the offset
18732 field type, which is "pointer" mode in emutls.c.
18734 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18737 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18739 PR middle-end/81207
18740 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18741 gimple_vuse copying separately from gimple_vdef copying.
18743 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18745 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18748 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18749 Olivier Hainque <hainque@adacore.com>
18751 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18752 for all vxworks7 targets.
18753 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18754 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18755 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18756 variations for VX6/VX7 and 32/64bits later on in ...
18757 (VXWORKS_LIB_SPEC): Leverage new macros.
18758 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18759 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18761 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18763 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18764 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18766 2017-06-26 Carl Love <cel@us.ibm.com>
18768 * config/rs6000/rs6000-c.c: Add support for built-in functions
18769 vector bool char vec_reve (vector bool char);
18770 vector signed char vec_reve (vector signed char);
18771 vector unsigned char vec_reve (vector unsigned char);
18772 vector bool int vec_reve (vector bool int);
18773 vector signed int vec_reve (vector signed int);
18774 vector unsigned int vec_reve (vector unsigned int);
18775 vector bool long long vec_reve (vector bool long long);
18776 vector signed long long vec_reve (vector signed long long);
18777 vector unsigned long long vec_reve (vector unsigned long long);
18778 vector bool short vec_reve (vector bool short);
18779 vector signed short vec_reve (vector signed short);
18780 vector double vec_reve (vector double);
18781 vector float vec_reve (vector float);
18782 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18783 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18784 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18785 (altivec_vreve): New pattern.
18786 * config/rs6000/altivec.h (vec_reve): New define.
18787 * doc/extend.texi (vec_rev): Update the built-in documentation file
18788 for the new built-in functions.
18790 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18792 PR tree-optimization/71815
18793 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18795 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18797 (slsr_process_phi): Likewise.
18798 (replace_uncond_cands_and_profitable_phis): Don't replace a
18799 multiply candidate with a stride of 1 (copy or cast).
18800 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18802 (lowest_cost_path): Likewise.
18803 (total_savings): Likewise.
18805 2017-06-26 Richard Biener <rguenther@suse.de>
18808 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18809 Use def_builtin_pure for all gather builtins.
18811 2017-06-26 Richard Biener <rguenther@suse.de>
18813 PR tree-optimization/81203
18814 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18815 non-dominating BBs.
18817 2017-06-26 Marek Polacek <polacek@redhat.com>
18820 * doc/invoke.texi: Document -Wmultistatement-macros.
18822 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18824 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18825 arm_neon_ok_no_float_abi effective target.
18827 2017-06-26 Richard Biener <rguenther@suse.de>
18829 PR tree-optimization/80928
18830 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18831 (copy_bbs): Set BB_DUPLICATED flag early.
18832 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18834 (execute_on_shrinking_pred): Likewise.
18835 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18836 BB_DUPLICATED blocks.
18837 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18838 iterate over all PHIs considering removal of *gsi.
18840 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18842 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18845 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18847 * config/rs6000/rs6000-string.c: (expand_block_clear,
18848 do_load_for_compare, select_block_compare_mode,
18849 compute_current_alignment, expand_block_compare,
18850 expand_strncmp_align_check, expand_strn_compare,
18851 expand_block_move, rs6000_output_load_multiple)
18852 Move functions related to string/block move/compare
18853 to a separate file.
18854 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18855 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18856 for this function which is now used in two files.
18857 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18858 * config.gcc: Add rs6000-string.o to extra_objs for
18859 targets powerpc*-*-* and rs6000*-*-*.
18861 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18864 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18865 32-bit, since indexed is not valid for DImode.
18866 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18867 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18868 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18869 (define_peephole2 for Altivec d-form store): Likewise.
18872 * multiple_target.c (create_dispatcher_calls): Only create the
18873 dispatcher call if the function is the default clone of a
18874 versioned function.
18876 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18878 PR middle-end/80902
18879 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18880 a call, force the call to not be a tail call.
18882 2017-06-23 Jeff Law <law@redhat.com>
18884 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18887 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18889 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18890 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18891 handling for early expansion of vector shifts (sl,sr,sra,rl).
18892 (builtin_function_type): Add vector shift right instructions
18893 to the unsigned argument list.
18895 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18897 rtl-optimizatoin/79286
18898 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18899 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18900 trap. PIC register plus a const unspec without offset can never trap.
18902 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18904 * tree.h (builtin_structptr_type): New type.
18905 (builtin_structptr_types): Declare new array.
18906 * tree.c (builtin_structptr_types): New array.
18907 (free_lang_data, build_common_tree_nodes): Use it.
18909 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18912 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18915 2017-06-22 Matt Turner <mattst88@gmail.com>
18917 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18918 Lake models to skylake case. Assume skylake for unknown
18919 models with clflushopt.
18921 2017-06-22 Jeff Law <law@redhat.com>
18923 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18924 frame sizes that do not satisfy aarch64_uimm12_shift.
18926 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18928 * profile-count.h (apply_probability,
18929 apply_scale, probability_in): Fix checks for zero.
18931 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18933 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18934 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18936 2016-06-22 Richard Biener <rguenther@suse.de>
18938 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18939 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18940 REDUC_MAX_EXPR support.
18941 (vectorizable_reduction): Likewise.
18942 (vect_create_epilog_for_reduction): Likewise.
18944 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18946 * match.pd (A / (1 << B) -> A >> B): New.
18947 * generic-match-head.c: Include optabs-tree.h.
18948 * gimple-match-head.c: Likewise.
18949 * optabs-tree.h (target_supports_op_p): New.
18950 * optabs-tree.c (target_supports_op_p): New.
18952 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18954 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18955 $gcc_cv_ld --help output.
18956 (gcc_cv_ld_demangle): Likewise.
18957 (gcc_cv_ld_eh_frame_hdr): Likewise.
18958 (gcc_cv_ld_pie): Likewise.
18959 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18960 (gcc_cv_ld_buildid): Likewise.
18961 (gcc_cv_ld_sysroot): Likewise.
18962 (ld_bndplt_support): Likewise.
18963 (ld_pushpopstate_support): Likewise.
18964 * configure: Regenerate.
18965 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18967 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18970 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18971 operands indexes to avoid gap between operands and match_dups.
18973 2017-06-21 Andrew Pinski <apinski@cavium.com>
18975 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18976 Increment Arith_shift and Arith_shift_reg by 1.
18977 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18979 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18980 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18981 (aarch64_strip_extend): Add new argument and test for it.
18982 (aarch64_cheap_mult_shift_p): New function.
18983 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18984 add a cost if it is true.
18985 Update calls to aarch64_strip_extend.
18986 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18988 2017-06-21 Andrew Pinski <apinski@cavium.com>
18990 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18992 (thunderxt88): Likewise.
18993 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18994 (thunderx_prefetch_tune): New variable.
18995 (thunderx2t99_prefetch_tune): Update for the correct values.
18996 (thunderxt88_tunings): New variable.
18997 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18998 generic_prefetch_tune.
18999 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19001 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19003 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19004 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19005 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19006 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19007 (aarch64_atomic_cas<mode>, GPI): Likewise.
19009 2017-06-21 Martin Liska <mliska@suse.cz>
19011 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19012 statements on cold and hot labels.
19013 * predict.c (tree_estimate_probability_bb): Remove the
19014 prediction from this place.
19016 2017-06-21 Martin Liska <mliska@suse.cz>
19018 PR tree-optimization/79489
19019 * gimplify.c (maybe_add_early_return_predict_stmt): New
19021 (gimplify_return_expr): Call the function.
19022 * predict.c (tree_estimate_probability_bb): Remove handling
19024 * predict.def: Update comment about early return predictor.
19025 * gimple-predict.h (is_gimple_predict): New function.
19026 * predict.def: Change default value of early return to 66.
19027 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19029 * passes.def: Put pass_strip_predict_hints to the beginning of
19032 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19034 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19035 FUNCTION_DECL declarations.
19036 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19038 (dwaf2out_decl): Likewise.
19039 * godump.c (go_early_global_decl): Skip call to the real debug hook
19040 for FUNCTION_DECL declarations.
19041 * passes.c (rest_of_decl_compilation): Skip call to the
19042 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19043 -fdump-go-spec is passed.
19045 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19047 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19048 Reorder for compactness.
19049 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19050 (def_builtin_pure, def_builtin_pure2): New functions.
19051 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19053 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19055 * match.pd (nop_convert): New predicate.
19056 ((A +- CST1) +- CST2): Allow some NOP conversions.
19058 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19061 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19062 with ctors/dtors if GOVD_SHARED is set.
19064 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19066 * config/aarch64/aarch64.md (movti_aarch64):
19067 Emit mov rather than orr.
19068 (movtf_aarch64): Likewise.
19069 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19070 Emit mov rather than orr.
19072 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19074 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19075 Swap alternatives, make integer dup more expensive.
19077 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19079 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19080 Return true for non-tls symbols.
19082 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19084 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19085 (cortex-a75): Likewise.
19086 (cortex-a75.cortex-a55): Likewise.
19087 * config/aarch64/aarch64-tune.md: Regenerate.
19088 * doc/invoke.texi (-mtune): Document new values for -mtune.
19090 2017-06-21 Tom de Vries <tom@codesourcery.com>
19092 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19093 stack_size feature.
19094 (Effective-Target Keywords, Other attributes): Suggest using
19095 dg-add-options stack_size feature to get stack limit in stack_size
19096 effective target documentation.
19098 2017-06-21 Julian Brown <julian@codesourcery.com>
19099 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19101 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19102 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19103 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19105 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19106 attribute type list for neon_multiply.
19107 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19108 attribute type list for neon_multiply.
19109 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19110 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19111 attribute type list for neon_multiply.
19112 * config/arm/types.md (crypto_pmull): Add.
19113 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19114 attribute type list.
19116 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19118 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19121 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19123 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19124 to make sure not to dereference a NULL cost_classes_ptr pointer.
19126 2017-06-20 Carl Love <cel@us.ibm.com>
19128 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19129 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19130 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19131 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19132 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19133 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19134 VMULOSW): New enum "unspec" values.
19135 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19136 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19137 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19138 altivec_vmulosw): New patterns.
19139 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19140 VMULOSW): Add definitions.
19142 2017-06-20 Julia Koval <julia.koval@intel.com>
19144 * config/i386/i386.c: Fix rounding expand for new pattern.
19145 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19147 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19149 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19150 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19152 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19154 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19157 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19159 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19160 architecture, then by implementer ID.
19161 * config/aarch64/aarch64-tune.md: Regenerate.
19163 2017-06-20 Richard Biener <rguenther@suse.de>
19165 PR middle-end/81097
19166 * fold-const.c (split_tree): Fold to type before negating.
19168 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19170 * diagnostic-show-locus.c
19171 (selftest::test_fixit_deletion_affecting_newline): New function.
19172 (selftest::diagnostic_show_locus_c_tests): Call it.
19174 2017-06-20 Andreas Schwab <schwab@suse.de>
19177 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19180 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19182 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19183 __ARM_FEATURE_COPROC according to support.
19185 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19187 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19188 Rewritten to avoid overflow for > 32-bit pointers.
19191 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19192 by removing enum keyword.
19193 (ubsan_type_descriptor): Likewise. Formatting fix.
19196 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19197 splitter): Require TARGET_SSE2 in the condition.
19199 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19202 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19203 for doing vector set of SFmode on ISA 3.0.
19204 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19205 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19207 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19208 SFmode value into a V4SF variable that was extracted from another
19209 V4SF variable without converting the element to double precision
19210 and back to single precision vector format.
19211 (vsx_insert_extract_v4sf_p9_2): Likewise.
19213 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19215 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19216 in UWHI to avoid undefined overflow.
19219 * ubsan.h (enum ubsan_encode_value_phase): New.
19220 (ubsan_encode_value): Change second argument to
19221 enum ubsan_encode_value_phase with default value of
19222 UBSAN_ENCODE_VALUE_GENERIC.
19223 * ubsan.c (ubsan_encode_value): Change second argument to
19224 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19225 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19226 create_tmp_var_raw instead of create_tmp_var and use a
19228 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19229 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19230 ubsan_encode_value callers.
19233 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19234 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19235 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19237 2017-06-19 Richard Biener <rguenther@suse.de>
19239 PR middle-end/81118
19240 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19241 estimates if we changed anything.
19243 2017-06-19 Richard Biener <rguenther@suse.de>
19245 PR tree-optimization/80887
19246 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19247 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19248 simplified lookups, then reset mprts_hook.
19249 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19251 (try_to_simplify): Likewise.
19253 2017-06-19 Martin Liska <mliska@suse.cz>
19256 * gimplify.c (gimplify_switch_expr):
19257 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19259 2017-06-19 Martin Liska <mliska@suse.cz>
19261 * doc/install.texi: Document that PGO runs in 4 stages.
19263 2017-06-19 Martin Liska <mliska@suse.cz>
19266 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19267 to dispatcher function name.
19268 * multiple_target.c (replace_function_decl): New function.
19269 (create_dispatcher_calls): Redirect both edges and references.
19271 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19273 * profile-count.c (profile_count::dump): Dump quality.
19274 (profile_count::differs_from_p): Update for unsigned val.
19275 * profile-count.h (profile_count_quality): New enum.
19276 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19278 2017-06-19 Richard Biener <rguenther@suse.de>
19280 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19281 struct function as arg.
19282 (estimate_numbers_of_iterations): Export overload with loop arg.
19283 (free_numbers_of_iterations_estimates_loop): Use an overload of
19284 free_numbers_of_iterations_estimates instead.
19285 * tree-cfg.c (remove_bb): Adjust.
19286 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19287 * tree-parloops.c (gen_parallel_loop): Likewise.
19288 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19290 (tree_unroll_loops_completely): Likewise.
19291 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19292 Use an overload instead and export.
19293 (estimated_loop_iterations): Adjust.
19294 (max_loop_iterations): Likewise.
19295 (likely_max_loop_iterations): Likewise.
19296 (estimate_numbers_of_iterations): Take struct function as arg
19298 (loop_exits_before_overflow): Adjust.
19299 (free_numbers_of_iterations_estimates_loop): Use an overload.
19300 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19301 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19303 2017-06-19 Richard Biener <rguenther@suse.de>
19306 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19307 RANGE_EXPR conservatively.
19309 2017-06-16 Carl Love <cel@us.ibm.com>
19311 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19312 definitions for vec_float, vec_float2, vec_floato,
19313 vec_floate built-ins.
19314 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19315 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19317 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19318 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19319 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19320 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19321 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19322 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19323 vec_floato): Add builtin defines.
19324 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19325 Update the built-in documentation file for the new built-in
19328 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19330 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19331 (mthumb): Mark as the negative of -marm.
19333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19335 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19337 (ARM Options, -mtune): Document that this accepts the same
19338 extension options as -mcpu.
19339 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19343 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19344 permitted extensions.
19346 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19348 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19349 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19350 (armv8-m.main): Add option +nodsp.
19351 * config/arm/arm-cpu-cdata.h: Regenerated.
19353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19355 * config/arm/t-fuchsia: New file.
19356 * config.gcc (arm*-*-fuchsia*): Use it.
19358 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19360 * config/arm/t-symbian: Rewrite for new option infrastructure.
19362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19364 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19365 (MULTILIB_REQUIRED): Likewise.
19367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19369 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19370 (MULTILIB_RESUE): Likewise.
19371 (MULTILIB_MATCHES): Likewise.
19372 (MULTLIB_REQUIRED): Likewise.
19374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19376 * config/arm/t-rtems: Rewrite for new option framework.
19378 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19380 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19381 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19382 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19383 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19384 * config/arm/t-multilib: ... here.
19385 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19386 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19387 armv7-a and armv8*-a when A-profile libraries have not been built.
19388 * config/arm/t-rmprofile: Rewrite.
19390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19392 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19393 with a backslash. Remove the backslash after substituting unescaped
19395 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19399 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19400 the driver to pass through the default setting of -mfloat-abi.
19401 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19403 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19404 (all_feat_combs): New rule.
19405 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19407 * config/arm/t-aprofile: Rewrite.
19409 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19411 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19412 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19413 fpu is not specified by the user/command-line.
19414 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19415 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19416 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19417 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19418 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19419 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19423 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19424 * config/arm/t-arm-elf: Rewritten.
19426 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19428 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19429 have some floating-point instructions.
19430 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19431 (TARGET_MAYBE_HARD_FLOAT): New macro.
19432 * config/arm/arm-builtins.c (arm_init_builtins): Use
19433 TARGET_MAYBE_HARD_FLOAT.
19434 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19436 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19438 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19439 (configargs.h): Include it.
19440 (arm_print_hint_for_fpu_option): New function.
19441 (arm_parse_fpu_option): New function.
19442 (candidate_extension): New class.
19443 (arm_canon_for_multilib): New function.
19444 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19445 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19446 (ARCH_CANONICAL_SPECS): New macro.
19447 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19451 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19452 are set after handling multilib fragments. Set target_cpu_default2
19455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19457 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19459 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19460 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19462 (check_arch): Likewise.
19463 * config/arm/arm.c (arm_configure_build_target): Handle
19464 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19465 options in the default.
19467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19469 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19470 when an option is an alias of another.
19471 * config/arm/parsecpu.awk (optalias): New parser token.
19472 (gen_comm_data): Mark non-alias options as such. Emit entries
19473 for extension aliases.
19474 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19475 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19476 (armv6kz, armv6zk, armv6t2): Likewise.
19477 (armv7): Make vfpv3-d16 an alias.
19478 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19480 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19481 Sort in canonical order.
19482 (armv8-a): Sort in canonical order.
19483 (armv8.1-a, armv8.2-a): Likewise.
19484 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19486 (cortex-a9): Sort in canonical order.
19487 * config/arm/arm.c (selftests.h): Include it.
19488 (arm_test_cpu_arch_data): New function.
19489 (arm_run_self_tests): New function.
19490 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19491 (targetm): Move declaration to the end of the file.
19492 * arm-cpu-cdata.h: Regenerated.
19494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19496 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19497 call to target_mode_check describing the type of option passed.
19498 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19499 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19500 arm_parse_cpu_option_name to match parameters against list of
19502 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19503 arm_arch_core_flags data structure.
19504 * config/arm/arm-cpu_cdata.h: Regenerated.
19506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19508 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19510 (arm_print_hint_for_cpu_option): Likewise.
19511 (arm_print_hint_for_arch_option): Likewise.
19512 (arm_parse_cpu_option_name): Likewise.
19513 (arm_parse_arch_option_name): Likewise.
19514 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19515 of entries in the all_fpus list.
19516 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19517 (arm_parse_cpu_option_name): Declare.
19518 (arm_parse_arch_option_name): Declare.
19519 (arm_parse_option_features): Declare.
19520 (arm_intialize_isa): Declare.
19521 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19523 (gen_comm_data): ... here. Make definitions non-static.
19524 * config/arm/arm-cpu-data.h: Regenerated.
19525 * config/arm/arm-cpu-cdata.h: Regenerated.
19527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19529 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19530 (cpu_arch_extension): New structure.
19531 (cpu_arch_option, arch_option, cpu_option): New structures.
19532 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19533 architecture types.
19534 (gen_data): Generate new format data tables.
19535 * config/arm/arm.c (cpu_tune): New structure.
19536 (cpu_option, processors): Delete.
19537 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19538 (arm_print_hint_for_cpu_option): ... this and ...
19539 (arm_print_hint_for_arch_option): ... this.
19540 (arm_parse_arch_cpu_name): Delete. Replace with ...
19541 (arm_parse_cpu_option_name): ... this and ...
19542 (arm_parse_arch_option_name): ... this.
19543 (arm_unrecognized_feature): Change type of target parameter to
19545 (arm_parse_arch_cpu_features): Delete. Replace with ...
19546 (arm_parse_option_features): ... this.
19547 (arm_configure_build_target): Rework to use new configuration data
19549 (arm_print_tune_info): Rework for new configuration data tables.
19550 * config/arm/arm-cpu-data.h: Regenerated.
19551 * config/arm/arm-cpu.h: Regenerated.
19553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19555 * Makefile.in (OBJS): Move sbitmap.o from here ...
19556 (OBJS-libcommon): ... to here.
19558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19560 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19561 (ISA_ALL_CRYPTO): New macro.
19562 (ISA_ALL_SIMD): New macro
19563 (ISA_ALL_FP): New macro.
19564 * config/arm/arm.c (fpu_bitlist): Update initializer.
19565 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19567 (arm9e): Add fpu. Add option for nofp
19568 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19569 (arm926ej-s, arm1026ej-s): Likewise.
19570 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19571 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19572 neon-fp16, neon-vfpv4, nofp and nosimd.
19573 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19574 (cortex-a8): Add fpu. Add option for nofp.
19575 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19576 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19577 (cortex-r4f): Add fpu.
19578 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19579 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19581 (cortex-r8): Likewise.
19582 (cortex-m4): Add fpu. Add option for nofp.
19583 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19584 (cortex-a17.cortex-a7): Likewise.
19585 (cortex-a32): Add fpu. Add options for crypto and nofp.
19586 (cortex-a35, cortex-a53): Likewise.
19587 (cortex-a57): Add fpu. Add option for crypto.
19588 (cortex-a72, cortex-a73): Likewise.
19589 (exynos-m1): Likewise.
19590 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19591 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19592 (cortex-m33): Add fpu. Add option for nofp.
19593 * config/arm/arm-cpu-cdata.h: Regenerated
19594 * config/arm/arm-cpu-data.h: Regenerated.
19596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19598 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19599 (armv5te, armv5tej): Likewise.
19600 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19601 (armv7): Add options fp and vfpv3-d16.
19602 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19603 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19605 (armv7ve): Likewise.
19606 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19607 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19608 (armv8-a): Add nocrypto option.
19609 (armv8.1-a, armv8.2-a): Likewise.
19610 (armv8-m.main): add options fp, fp.dp and nofp.
19612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19614 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19616 (armv8-a+crc): Delete.
19617 (armv8.1-a): Add options simd, crypto and nofp.
19618 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19619 (armv8.2-a+fp16): Delete.
19620 (armv8-m.main): Add option dsp.
19621 (armv8-m.main+dsp): Delete.
19622 (cortex-a8): Add fpu. Add nofp option.
19623 (cortex-a9): Add fpu. Add nofp and nosimd options.
19624 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19625 link to main cpu and architecture data structures.
19626 (gen_comm_data): Only put isa attributes from the main architecture
19628 (option): New statement for architecture and CPU entries.
19629 * arm.c (struct cpu_option): New structure.
19630 (struct processors): Add entry for options.
19631 (arm_unrecognized_feature): New function.
19632 (arm_parse_arch_cpu_name): Ignore any characters after the first
19634 (arm_parse_arch_cpu_feature): New function.
19635 (arm_configure_build_target): Separate out any CPU and architecture
19636 features and parse separately. Don't error out if -mfpu=auto is
19637 used with only an architecture string.
19638 (arm_print_asm_arch_directives): New function.
19639 (arm_file_start): Call it.
19640 * config/arm/arm-cpu-cdata.h: Regenerated.
19641 * config/arm/arm-cpu-data.h: Likewise.
19642 * config/arm/arm-tables.opt: Likewise.
19644 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19646 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19647 assembler when it is not -mfpu=auto.
19649 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19651 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19652 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19653 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19654 (ASM_CPU_SPEC): Rewrite.
19655 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19656 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19657 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19658 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19659 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19660 copied string is NUL-terminated. Also strip any characters prefixed
19662 (arm_rewrite_selected_arch): New function.
19663 (arm_rewrite_march): New function.
19665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19667 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19668 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19669 (march, mcpu, mtune): Convert to string-based options.
19670 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19671 (arm_parse_arch_cpu_name): New function.
19672 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19673 identify selected architecture or CPU.
19674 (arm_option_save): New function.
19675 (TARGET_OPTION_SAVE): Redefine.
19676 (arm_option_restore): Restore string options.
19677 (arm_option_print): Print string options.
19679 2017-06-16 Martin Sebor <msebor@redhat.com>
19681 PR tree-optimization/80933
19682 PR tree-optimization/80934
19683 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19684 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19685 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19686 (gimple_fold_builtin): Call them.
19688 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19690 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19691 as unlikely; update profile.
19693 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19695 * predict.c (force_edge_cold): Handle declaring edges impossible
19698 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19700 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19702 (try_unroll_loop_completely): Fix reporting.
19704 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19706 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19708 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19711 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19712 if given a non-constant argument for an intrinsic which requires a
19715 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19717 * profile.c (compare_freqs): New function.
19718 (branch_prob): Sort edge list.
19719 (find_spanning_tree): Assume that the list is priority sorted.
19721 2017-06-16 Richard Biener <rguenther@suse.de>
19723 PR tree-optimization/81090
19724 * passes.def (pass_record_bounds): Remove.
19725 * tree-pass.h (make_pass_record_bounds): Likewise.
19726 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19727 make_pass_record_bounds): Likewise.
19728 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19729 not free niter estimates at the beginning but at the end.
19730 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19732 2017-06-16 Richard Biener <rguenther@suse.de>
19734 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19735 initializer to workaround ICE in host GCC 4.8.
19737 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19739 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19741 (clone_inlined_nodes): Update.
19743 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19745 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19746 prefetch settings, and enable prefetching by default at -O3.
19748 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19750 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19751 Set flag_prefetch_loop_arrays according to tuning data.
19753 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19755 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19756 New tune structure.
19757 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19758 [Unrelated to main purpose of the patch] Place the pointer field last
19759 to enable type checking errors when tune structure are wrongly merged.
19760 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19761 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19762 (thunderx2t99_prefetch_tune): New tune constants.
19763 (tune_params *_tunings): Update all tunings (no functional change).
19764 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19765 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19766 from tunings structures.
19768 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19771 * ubsan.c (instrument_null): Add T argument, use it instead
19772 of computing it based on IS_LHS.
19773 (instrument_object_size): Likewise.
19774 (pass_ubsan::execute): Adjust instrument_null and
19775 instrument_object_size callers to pass gimple_get_lhs or
19776 gimple_assign_rhs1 result to it. Use instrument_null instead of
19777 calling get_base_address and instrument_mem_ref. Handle
19778 aggregate call arguments for object-size sanitization.
19780 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19782 PR tree-optimization/81089
19783 * tree-vrp.c (is_masked_range_test): Validate operands of
19786 2017-06-15 Martin Sebor <msebor@redhat.com>
19789 * dumpfile.c (dump_register): Avoid calling memset to initialize
19790 a class with a default ctor.
19791 * gcc.c (struct compiler): Remove const qualification.
19792 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19793 * hash-table.h: Ditto.
19794 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19796 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19797 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19799 * params.h (struct param_info): Make struct members non-const.
19800 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19801 with default initialization.
19802 * vec.h (vec_copy_construct, vec_default_construct): New helper
19804 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19805 with vec_copy_construct.
19806 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19807 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19808 * doc/invoke.texi (-Wclass-memaccess): Document.
19810 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19812 * emit-rtl.h (is_leaf): Update comment about local
19813 register allocator.
19815 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19818 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19819 for a variable to have a section before checking if the section has a
19821 Set section to.persistent if persistent attribute is set.
19822 Warn if .persistent attribute is used on an automatic variable.
19824 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19826 PR rtl-optimization/80474
19827 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19829 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19831 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19834 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19836 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19837 (casesi): Emit bounds checking as RTL.
19838 (casesi_internal_mips16_<mode>): Remove bounds checking.
19840 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19842 * config/xtensa/xtensa.c (xtensa_option_override): Append
19843 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19844 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19845 xtensa_doloop_hooks): Define unconditionally.
19846 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19848 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19849 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19850 for it in xtensa_option_override.
19851 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19852 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19854 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19856 * doc/cppopts.texi: Document '-' special value to -MF.
19858 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19860 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19861 (cortex_a53_fconst): Likewise.
19862 (cortex_a53_fpmul): Likewise.
19863 (cortex_a53_f_load_64): Likewise.
19864 (cortex_a53_f_load_many): Likewise.
19865 (cortex_a53_advsimd_alu): Likewise.
19866 (cortex_a53_advsimd_alu_q): Likewise.
19867 (cortex_a53_advsimd_mul): Likewise.
19868 (cortex_a53_advsimd_mul_q): Likewise.
19869 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19870 Add missing fmul, r2f_cvt and fconst cases.
19872 2017-06-14 Richard Biener <rguenther@suse.de>
19874 PR middle-end/81088
19875 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19877 (fold_binary_loc): When associating do not treat pre-existing
19878 TREE_OVERFLOW on literal constants as a reason to allow
19879 TREE_OVERFLOW on associated literal constants.
19881 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19883 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19884 (MASK_FEATURES): New macro.
19885 * config/sparc/sparc.c (sparc_option_override): Remove the special
19886 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19888 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19890 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19891 a division of 0 if non-call exceptions are enabled.
19893 2017-06-14 Andrew Pinski <apinski@cavium.com>
19894 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19897 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19898 Improve vector initialization code gen for only variable case.
19900 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19902 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19904 2017-06-14 Richard Biener <rguenther@suse.de>
19906 PR tree-optimization/81083
19907 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19910 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19912 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19913 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19914 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19915 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19916 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19917 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19919 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19921 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19922 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19924 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19926 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19928 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19930 * config/rs6000/t-rtems: Don't handle SPE.
19932 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19934 * config/rs6000/t-linux: Don't handle SPE.
19936 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19938 * config/rs6000/eabispe.h: Delete file.
19940 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19942 * config/rs6000/t-spe: Delete file.
19944 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19946 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19947 (rs6000_legitimate_offset_address_p): Return false for anything in
19948 V2SImode or V2SFmode.
19950 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19952 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19953 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19954 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19956 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19957 (rs6000_legitimate_offset_address_p): Ditto.
19958 (rs6000_emit_move): Ditto.
19959 (rs6000_init_builtins): Remove V4HI_type_node.
19961 2017-06-13 Martin Liska <mliska@suse.cz>
19964 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19965 (gate_asan): Likewise.
19966 * asan.h (asan_no_sanitize_address_p): Remove the function.
19967 (sanitize_flags_p): New function.
19968 * builtins.def: Fix coding style.
19969 * common.opt: Use renamed enum value.
19970 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19971 * doc/extend.texi: Document no_sanitize attribute.
19972 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19973 to SANITIZE_UNDEFINED_NONDEFAULT.
19974 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19975 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19976 Use sanitize_flags_p.
19977 * gimplify.c (gimplify_function_tree): Likewise.
19978 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19979 * opts.c (parse_no_sanitize_attribute): New function.
19980 (common_handle_option): Use renamed enum value.
19981 * opts.h (parse_no_sanitize_attribute): Declare.
19982 * tree.c (sanitize_flags_p): New function.
19983 * tree.h: Declared here.
19984 * tsan.c: Use sanitize_flags_p.
19985 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19986 (instrument_mem_ref): Likewise.
19987 (instrument_bool_enum_load): Likewise.
19988 (do_ubsan_in_current_function): Remove the function.
19989 (pass_ubsan::execute): Use sanitize_flags_p.
19990 * ubsan.h: Remove do_ubsan_in_current_function
19991 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19992 (dump_function_to_file): Use it here.
19994 2017-06-13 Martin Jambor <mjambor@suse.cz>
19996 PR tree-optimization/80803
19997 PR tree-optimization/81063
19998 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19999 (propagate_subaccesses_across_link): Enqueue subtree whenever
20000 necessary instead of relying on the caller.
20002 2017-06-13 Martin Jambor <mjambor@suse.cz>
20004 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20005 that have a first_link.
20006 (sort_and_splice_var_accesses): Do not check first_link before
20008 (subtree_mark_written_and_enqueue): Likewise.
20009 (propagate_all_subaccesses): Likewise and do not stop at first
20010 parent with a first_link.
20012 2017-06-13 Martin Jambor <mjambor@suse.cz>
20014 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20017 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20019 * match.pd: New pattern.
20021 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20023 * tree-vrp.c (is_masked_range_test): New function.
20024 (register_edge_assert_for): Determine ranges for
20027 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20029 PR tree-optimization/67328
20030 * fold-const.c (maskable_range_p): New function.
20031 (build_range_check): Generate bittests if possible.
20033 2017-06-13 Martin Liska <mliska@suse.cz>
20035 * gimple-pretty-print.c (dump_probability): Add new argument.
20036 (dump_edge_probability): Dump both probability and count.
20037 (dump_gimple_label): Likewise.
20038 (dump_gimple_bb_header): Likewise.
20040 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20043 * config/avr/avr-devices.c: Fix indentation.
20044 * config/avr/gen-avr-mmcu-specs.c: Dito.
20046 2017-06-13 Richard Biener <rguenther@suse.de>
20048 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20049 instead get vector type from stmt_info.
20050 (vectorizable_reduction): Adjust. Remove dead code.
20052 2017-06-13 Richard Biener <rguenther@suse.de>
20054 PR middle-end/81065
20055 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20056 case of C * (x * C2 + C3).
20057 (fold_addr_of_array_ref_difference): Properly fold index difference.
20059 2017-06-12 David S. Miller <davem@davemloft.net>
20062 * config/sparc/sparc.md (return expander): Emit frame blockage if
20063 function uses alloca.
20065 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20067 * combine.c (make_field_assignment): Check len rather than the mode
20068 precision when calling force_to_mode.
20070 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20072 Support multilibs and devices that see flash in RAM address range.
20075 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20076 (avr_mcu_t) <flash_pm_offset>: New field.
20077 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20078 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20079 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20080 (AVR_TINY_PM_OFFSET): Remove macro.
20081 * config/avr/avr.opt (-mshort-calls): New option.
20082 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20083 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20084 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20085 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20086 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20087 instead of avr_arch->have_jmp_call.
20088 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20089 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20090 avr_arch->flash_pm_offset to define.
20091 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20092 new field flash_pm_offset. Add entry for avrxmega3.
20093 (avr_texinfo): Add entry for avrxmega3.
20094 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20095 attiny212, attiny214,
20096 attiny412, attiny414, attiny416, attiny417,
20097 attiny814, attiny816, attiny817,
20098 attiny1614, attiny1616, attiny1617,
20099 attiny3214, attiny3216, attiny3217.
20100 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20101 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20102 (avr_print_operand_address) [AVR_TINY]: Same.
20103 (avr_asm_init_sections) <readonly_data_section>: Only patch
20104 callback if avr_arch->flash_pm_offset = 0.
20105 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20106 for rodata if avr_arch->flash_pm_offset != 0.
20107 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20108 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20109 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20110 (m_options): Append opt_rcall.
20111 (m_dirnames): Append dir_rcall.
20112 * config/avr/t-multilib: Regenerate.
20114 * configure.ac [target=avr]: Check whether avrxmega3 default
20115 linker description file works as needed.
20116 * configure: Regenerate.
20117 * doc/avr-mmcu.texi: Regenerate.
20118 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20119 <__AVR_ARCH__>: Document avrxmega3 and 103.
20120 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20121 <__AVR_SHORT_CALLS__>: Document it.
20122 <__AVR_PM_BASE_ADDRESS__>: Document it.
20123 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20124 (AVR Variable Attributes) <progmem>: Document this is
20125 not needed for avrxmega3.
20126 (AVR Named Address Spaces) <__flash>: Dito.
20128 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20130 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20132 2017-06-12 Doug Rupp <rupp@adacore.com>
20134 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20135 Append vxworks-stdint.h to the tm_file list.
20136 * config/vxworks-stdint.h: New file.
20138 2017-06-12 Martin Liska <mliska@suse.cz>
20140 PR tree-optimization/81041
20141 * tree-profile.c (gimple_gen_ic_func_profiler):
20142 Create an extra BB in profile-generate
20143 (gimple_gen_time_profiler): Likewise.
20145 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20147 PR tree-optimization/81003
20148 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20149 (update_range_test): Use it instead of force_gimple_operand_gsi.
20151 2017-06-12 Richard Biener <rguenther@suse.de>
20153 PR tree-optimization/81053
20154 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20155 with backedge value not defined in loop. Simplify def stmt
20158 2017-06-11 Tom de Vries <tom@codesourcery.com>
20161 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20163 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20164 nvptx_cannot_force_const_mem.
20166 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20168 * opts.c (finish_options): Move test for flag_split_stack after
20169 it has been initialized.
20171 2017-06-11 Jason Merrill <jason@redhat.com>
20173 * tree.h (id_equal): New.
20174 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20175 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20176 instead of strcmp of IDENTIFIER_POINTER.
20178 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20180 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20181 (mark_all_inlined_calls_cdtor): Fix formating.
20182 (inline_transform): Rescale profile before inlining.
20184 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20186 * cgraph.h (cgraph_edge::clone): Update prototype.
20187 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20188 (cgraph_node::create_clone): Update.
20189 (cgraph_node::create_version_clone): Update.
20190 * tree-inline.c (copy_bb): Update.
20191 (expand_call_inline): Update.
20193 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20195 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20196 factored out from ...
20197 (rs6000_emit_prologue): ... here.
20199 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20201 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20202 factored out from ...
20203 (rs6000_emit_prologue): ... here.
20205 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20207 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20209 (handle_missing_profiles): Fix computation of tp_first_run.
20210 (counts_to_freqs): Do not touch freqs when count is 0.
20212 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20214 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20217 2017-06-10 Tom de Vries <tom@codesourcery.com>
20219 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20220 attributes): Document signal effective target.
20222 2017-06-10 Tom de Vries <tom@codesourcery.com>
20224 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20225 Document effective target stack_size.
20227 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20229 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20230 to the edit_context if they can be auto-applied.
20232 2017-06-9 Ian Lance Taylor <iant@golang.org>
20234 * opts.c (finish_options): If -fsplit-stack, disable implicit
20235 -forder-blocks-and-partition.
20236 * doc/invoke.texi (Optimize Options): Document that when using
20237 -fsplit-stack -forder-blocks-and-partition is not implicitly
20240 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20242 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20243 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20244 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20245 * builtins.def (abort, trap, unreachable): Declare cold.
20246 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20247 * tree-core.h (ECF_COLD): New.
20248 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20249 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20251 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20253 * predict.c (unlikely_executed_stmt_p): Cleanup.
20255 2017-06-09 Richard Biener <rguenther@suse.de>
20257 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20258 model if the ref is always written to.
20260 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20262 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20264 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20266 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20269 2017-06-09 Tom de Vries <tom@codesourcery.com>
20272 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20273 "target cannot support label values" when encountering LABEL_REF.
20275 2017-06-09 Martin Liska <mliska@suse.cz>
20277 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20278 (gimple_gen_ic_func_profiler): Emit direct comparison
20279 of __gcov_indirect_call_callee with NULL.
20280 (gimple_gen_time_profiler): Change probability from
20281 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20283 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20285 * profile.c (edge_gcov_counts): Turn to pointer.
20286 (compute_branch_probabilities, compute_branch_probabilities): Update.
20287 (branch_prob): Do not clear edge_gcov_count.
20288 * profile.h (edge_gcov_counts): Turn to pointer.
20289 (edge_gcov_count): Update.
20291 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20293 * gimple.h (gimple_check_failed): Mark cold.
20295 2017-06-09 Richard Biener <rguenther@suse.de>
20297 PR tree-optimization/66623
20298 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20299 refactor check_reduction into two parts, properly computing
20300 whether we have to check reduction validity for outer loop
20303 2017-06-09 Richard Biener <rguenther@suse.de>
20305 PR tree-optimization/79483
20306 * graphite-scop-detection.c (order): New global.
20307 (get_order): Compute bb to order mapping that satisfies code
20308 generation constraints.
20309 (cmp_pbbs): New helper.
20310 (build_scops): Start domwalk at entry block, sort generated
20313 2017-06-09 Richard Biener <rguenther@suse.de>
20315 PR middle-end/81007
20316 * ipa-polymorphic-call.c
20317 (ipa_polymorphic_call_context::restrict_to_inner_class):
20318 Skip FIELD_DECLs with error_mark_node type.
20319 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20322 2017-06-09 Martin Liska <mliska@suse.cz>
20324 * predict.c (struct branch_predictor): New struct.
20325 (test_prediction_value_range): New test.
20326 (predict_c_tests): New function.
20327 * selftest-run-tests.c (selftest::run_tests): Run the function.
20328 * selftest.h: Declare new tests.
20330 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20333 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20334 gen_add3_insn did not fail.
20335 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20336 r0, construct that number in a temporary reg and add that reg to r0.
20337 If asked to put the result in r0 as well, fail.
20339 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20341 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20342 for early expansion of vec_eqv.
20344 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20346 PR middle-end/81005
20347 * ubsan.c (instrument_null): Avoid pointless code temporary.
20348 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20350 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20354 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20357 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20360 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20362 * predict.c (unlikely_executed_edge_p): Move ahead.
20363 (probably_never_executed_edge_p): Use it.
20365 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20367 PR middle-end/79988
20368 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20369 gimple_call_builtin_p call.
20371 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20373 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20374 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20375 rtl_check_failed_type2, rtl_check_failed_code1,
20376 rtl_check_failed_code2, rtl_check_failed_code_mode,
20377 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20378 rtvec_check_failed_bounds, rtl_check_failed_flag,
20379 _fatal_insn_not_found, _fatal_insn): Likewise.
20380 * tree.h (tree_contains_struct_check_failed,
20381 tree_check_failed, tree_not_check_failed,
20382 tree_class_check_failed, tree_range_check_failed,
20383 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20384 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20385 tree_operand_check_failed, omp_clause_check_failed,
20386 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20389 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20391 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20392 flag_branch_probabilities.
20393 * ipa-inline.c (edge_badness): Likewise.
20394 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20395 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20396 * predict.c (maybe_hot_frequency_p): Likewise.
20397 (probably_never_executed): Likewise.
20398 * sched-ebb.c (schedule_ebbs): Likewise.
20399 * sched-rgn.c (find_single_block_region): Likewise.
20400 * tracer.c (tail_duplicate): Likewise.
20402 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20404 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20405 longer requires x_flag_profile_use.
20407 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20409 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20410 instead of flag_reorder_blocks_and_partition.
20411 * dbxout.c (dbxout_function_end): Likewise.
20412 * dwarf2out.c (gen_subprogram_die): Likewise.
20413 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20414 * hw-doloop.c (reorg_loops): Likewise.
20415 * varasm.c (assemble_start_function,
20416 assemble_end_function): Likewise.
20417 (decide_function_section): Do not check for
20418 flag_reorder_blocks_and_partition.
20420 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20422 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20424 (chkp_get_hard_register_fake_addr_expr): Ditto.
20425 (chkp_build_addr_expr): Add check for hard reg case.
20426 (chkp_parse_array_and_component_ref): Ditto.
20427 (chkp_find_bounds_1): Ditto.
20428 (chkp_process_stmt): Don't generate bounds store for
20431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20433 * predict.c (maybe_hot_bb_p): Do not check profile status.
20434 (maybe_hot_edge_p): Likewise.
20435 (probably_never_executed): Check for zero counts even if profile
20437 (unlikely_executed_edge_p): New function.
20438 (unlikely_executed_stmt_p): New function.
20439 (unlikely_executed_bb_p): New function.
20440 (set_even_probabilities): Use unlikely predicates.
20441 (combine_predictions_for_bb): Likewise.
20442 (predict_paths_for_bb): Likewise.
20443 (predict_paths_leading_to_edge): Likewise.
20444 (determine_unlikely_bbs): New function.
20445 (estimate_bb_frequencies): Use it.
20446 (compute_function_frequency): Use zero counts even if profile is
20448 * profile-count.h: Fix typo.
20450 2017-08-08 Julia Koval <julia.koval@intel.com>
20452 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20453 _mm512_mask_cvtsepi16_storeu_epi8,
20454 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20455 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20456 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20457 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20458 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20459 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20460 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20461 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20462 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20463 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20464 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20465 __builtin_ia32_pmovuswb256mem_mask,
20466 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20467 __builtin_ia32_pmovwb512mem_mask): New builtins.
20469 2017-08-08 Julia Koval <julia.koval@intel.com>
20471 PR target/73350,80862
20472 * config/i386/subst.md (round): Fix round pattern.
20473 * config/i386/i386.c (ix86_erase_embedded_rounding):
20474 Fix erasing rounding for the fixed pattern.
20476 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20478 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20480 2017-06-08 Martin Liska <mliska@suse.cz>
20482 PR gcov-profile/80911
20483 * gcov.c (block_info::block_info): New constructor.
20485 2017-06-07 Carl Love <cel@us.ibm.com>
20487 * config/rs6000/rs6000-c: The return type of the following
20488 built-in functions was implemented as int not long long. Fix sign
20489 of return value for the unsigned version of vec_mulo and vec_mule.
20490 vector unsigned long long vec_bperm (vector unsigned long long,
20491 vector unsigned char)
20492 vector signed long long vec_mule (vector signed int,
20494 vector unsigned long long vec_mule (vector unsigned int,
20495 vector unsigned int)
20496 vector signed long long vec_mulo (vector signed int,
20498 vector unsigned long long vec_mulo (vector unsigned int,
20499 vector unsigned int)
20500 * doc/extend.texi: Fix the documentation for the built-in
20503 2017-06-07 Carl Love <cel@us.ibm.com>
20506 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20509 2017-06-07 Carl Love <cel@us.ibm.com>
20511 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20512 support, Generate doublehv for signed int/float for BE case only.
20514 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20516 * doc/invoke.texi (mcx16): Rewrite.
20518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20520 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20521 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20522 *mov<mode>_softfloat, and an anonymous splitter): Use
20523 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20525 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20527 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20529 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20530 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20531 (rs6000_debug_reg_global): Adjust.
20532 (rs6000_init_hard_regno_mode_ok): Adjust.
20533 (rs6000_dbx_register_number): Adjust.
20534 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20535 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20536 Remove SPE_ACC and SPEFSCR.
20537 (REG_ALLOC_ORDER): Ditto.
20538 (FRAME_POINTER_REGNUM): Change to 111.
20539 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20540 (REG_CLASS_NAMES): Ditto.
20541 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20542 (REGISTER_NAMES): Ditto.
20543 (ADDITIONAL_REG_NAMES): Ditto.
20544 (rs6000_reg_names): Ditto.
20545 * config/rs6000/rs6000.md: Renumber some register number
20548 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20550 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20552 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20553 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20555 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20556 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20557 Delete the SPE high registers.
20558 (REG_ALLOC_ORDER): Ditto.
20559 (enum reg_class): Remove SPE_HIGH_REGS.
20560 (REG_CLASS_NAMES): Ditto.
20561 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20562 (REGISTER_NAMES): Ditto.
20563 (rs6000_reg_names): Ditto.
20564 * doc/tm.texi.in: Remove SPE as example.
20565 * doc/tm.texi: Regenerate.
20567 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20569 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20570 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20571 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20572 * config/rs6000/rs6000.md (type): Remove "brinc".
20574 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20576 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20577 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20578 * config/rs6000/linuxspe.h: Delete file.
20579 * config/rs6000/rs6000.md: Don't include spe.md.
20580 * config/rs6000/spe.h: Delete file.
20581 * config/rs6000/spe.md: Delete file.
20582 * config/rs6000/t-rs6000: Remove spe.md.
20584 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20586 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20587 (reg_or_none500mem_operand): Delete.
20588 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20589 instead of reg_or_none500mem_operand.
20591 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20594 handling of SPE flags.
20595 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20597 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20599 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20601 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20602 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20603 paired_divv2sf3): Similar.
20604 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20605 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20606 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20608 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20609 Rename the paired_* instruction patterns.
20610 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20612 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20613 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20614 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20615 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20616 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20617 PAIRED_VECTOR_MODE.
20618 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20619 (spe_func_has_64bit_regs_p): Delete.
20620 (spe_expand_predicate_builtin): Delete.
20621 (spe_expand_evsel_builtin): Delete.
20622 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20623 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20624 (invalid_e500_subreg): Delete.
20625 (rs6000_legitimize_address): Always force_reg op2 as well, for
20626 paired single memory accesses.
20627 (rs6000_member_type_forces_blk): Delete.
20628 (rs6000_spe_function_arg): Delete.
20629 (rs6000_expand_unop_builtin): Delete SPE handling.
20630 (rs6000_expand_binop_builtin): Ditto.
20631 (spe_expand_stv_builtin): Delete.
20632 (bdesc_2arg_spe): Delete.
20633 (spe_expand_builtin): Delete.
20634 (spe_expand_predicate_builtin): Delete.
20635 (spe_expand_evsel_builtin): Delete.
20636 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20637 (spe_init_builtins): Delete.
20638 (spe_func_has_64bit_regs_p): Delete.
20639 (savres_routine_name): Delete "info" parameter. Adjust callers.
20640 (rs6000_emit_stack_reset): Ditto.
20641 (rs6000_dwarf_register_span): Delete.
20642 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20643 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20644 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20646 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20648 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20649 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20650 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20651 mulv2sf3, divv2sf3): Delete expanders.
20653 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20655 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20657 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20659 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20660 * config/rs6000/rs6000.c: Ditto.
20662 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20664 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20665 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20666 comparison_operator.
20668 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20670 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20671 * config/rs6000/rs6000.opt: Ditto.
20672 * config/rs6000/t-rtems: Ditto.
20674 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20676 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20677 TARGET_E500_SINGLE by 0, simplify.
20678 * config/rs6000/rs6000.c: Ditto.
20679 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20680 (spe_build_register_parallel): Delete.
20681 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20682 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20683 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20684 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20685 (E500_CONVERT): Delete.
20686 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20690 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20691 * config/rs6000/dfp.md: Ditto.
20692 (negdd2, *negdd2_fpr): Merge.
20693 (absdd2, *absdd2_fpr): Merge.
20694 (negtd2, *negtd2_fpr): Merge.
20695 (abstd2, *abstd2_fpr): Merge.
20696 * config/rs6000/e500.h: Delete file.
20697 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20698 TARGET_FPRS by 1 and simplify.
20699 * config/rs6000/rs6000-c.c: Ditto.
20700 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20701 TARGET_DF_SPE by 0.
20702 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20704 * config/rs6000/rs6000.md: Ditto.
20705 (floatdidf2, *floatdidf2_fpr): Merge.
20706 (move_from_CR_gt_bit): Delete.
20707 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20708 (E500_CR_IOR_COMPARE): Delete.
20709 (All patterns that require !TARGET_FPRS): Delete.
20710 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20712 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20714 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20716 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20718 * graphds.c (add_edge): Intitialize edge's attached data.
20719 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20720 pointer parameter. Call pointed function on each edge during
20721 graph traversing. Skip traversing the edge when the function
20723 (graphds_dfs, graphds_scc): Ditto.
20724 (for_each_edge): New parameter. Pass the new parameter to callback
20726 * graphds.h (skip_edge_callback): New function pointer type.
20727 (graphds_dfs, graphds_scc): New function pointer parameter.
20728 (graphds_edge_callback, for_each_edge): New parameter.
20730 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20732 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20733 out code checking if runtime alias check is possible to below ...
20734 Call the new function.
20735 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20736 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20738 2017-06-07 Marek Polacek <polacek@redhat.com>
20741 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20742 TYPE_OVERFLOW_WRAPS check.
20744 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20746 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20747 if versioning is required.
20748 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20749 peeling with the check for versioning.
20751 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20753 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20754 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20755 Set true to new parameter if new ssa variable is defined.
20756 (vect_gen_vector_loop_niters): Refactor. Set range information
20757 for the new vector loop bound variable.
20758 (vect_do_peeling): Ditto.
20760 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20762 * tree-affine.c (ssa.h): Include header file.
20763 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20764 has wrapping overflow behavior.
20766 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20768 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20770 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20772 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20773 (tree_to_aff_combination): ... here.
20775 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20777 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20778 reg_pressure model function.
20779 (ivopts_global_cost_for_size): Delete.
20780 (determine_set_costs, iv_ca_recount_cost): Call new model function
20781 ivopts_estimate_reg_pressure.
20783 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20785 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20786 expensive than udiv. Remove floating point cases from mod.
20788 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20790 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20791 Increase idiv cost.
20793 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20795 * config/aarch64/aarch64.md
20796 (copysignsf3): Fix mask generation.
20798 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20800 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20802 (class dump_manager): Add register_dumps method.
20803 * dumpfile.c: Include langhooks.h.
20804 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20805 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20806 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20807 (dump_manager::dump_register): Start with 512 entries instead of 32.
20808 (dump_manager::register_dumps): New method.
20809 * toplev.c (general_init): Instead of invoking register_dumps
20810 langhook, invoke register_dumps method on the dump manager.
20811 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20814 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20816 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20817 Say that # requires an associated define_split to exist, and that
20818 the define_split must be suitable for use after register allocation.
20820 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20822 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20823 (compute_outgoing_frequencies): Also initialize zero counts.
20824 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20825 around loops; preserve more of profile when nothing changes.
20827 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20829 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20831 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20832 * config/arm/arm-cpu-cdata.h: Regenerate.
20833 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20834 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20835 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20836 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20837 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20839 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20841 * config/arm/t-rmprofile: Likewise.
20842 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20844 2017-06-06 David S. Miller <davem@davemloft.net>
20847 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20848 blockage if function uses alloca.
20850 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20852 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20853 New "uid" fields to hold pretty-print IDs of group and ref.
20854 Memory references are now identified as <group_id>:<ref_id>
20855 instead of using [random] addresses.
20856 (dump_mem_details): Simplify, no functional change.
20857 (dump_mem_ref): Simplify and make output more concise.
20858 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20859 (find_or_create_group): Initialize group uid.
20860 (record_ref): Initialize ref uid. Improve debug output.
20861 (prune_group_by_reuse, should_issue_prefetch_p,)
20862 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20863 (mark_nontemporal_store, determine_loop_nest_reuse):
20864 Improve debug output.
20866 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20868 * dbgcnt.def (prefetch): New debug counter.
20869 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20870 (schedule_prefetches): Stop issueing prefetches if debug counter
20873 2017-06-06 Tom de Vries <tom@codesourcery.com>
20875 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20876 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20878 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20880 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20881 Use aarch64_reg_or_zero predicate for operand 4.
20882 (aarch64_compare_and_swap<mode> define_insn_and_split):
20883 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20884 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20886 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20888 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20889 (arm_compute_save_core_reg_mask): This.
20890 (thumb1_compute_save_reg_mask): Rename into ...
20891 (thumb1_compute_save_core_reg_mask): This.
20892 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20893 (arm_compute_frame_layout): Likewise.
20895 2017-06-06 Richard Biener <rguenther@suse.de>
20897 PR tree-optimization/80974
20898 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20899 keep or clear leaders SSA info.
20901 2017-06-06 Tom de Vries <tom@codesourcery.com>
20903 * config/nvptx/nvptx.c (split_mode_p): New function.
20904 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20906 2017-06-06 Tom de Vries <tom@codesourcery.com>
20908 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20910 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20913 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20916 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20918 * shrink-wrap.c (handle_simple_exit): Update profile.
20919 (try_shrink_wrapping): Upate profile.
20921 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20923 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20924 (tree_guess_outgoing_edge_probabilities): New.
20925 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20926 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20928 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20930 * ipa-split.c (split_function): Initialize return bb profile.
20932 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20934 * profile.c (compute_branch_probabilities): Also initialize
20935 EXIT_BLOCK profile.
20937 2017-06-06 Richard Biener <rguenther@suse.de>
20939 PR tree-optimization/80928
20940 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20941 (vect_analyze_loop_operations): Properly guard analysis for
20943 (vect_transform_loop): Likewise.
20944 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20945 (vect_model_induction_cost): Do not cost for pure SLP.
20946 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20947 of induction in inner loop vectorization.
20948 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20949 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20950 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20952 (vect_analyze_slp_cost_1): Cost induction.
20953 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20954 (vect_get_slp_vect_defs): Likewise.
20955 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20956 (vect_transform_stmt): Handle SLP reductions.
20957 * tree-vectorizer.h (vectorizable_induction): Adjust.
20959 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20961 * config/rs6000/rs6000.c (make_resolver_func): Update
20962 init_lowered_empty_function call.
20964 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20966 * doc/invoke.texi: Document the -fprofile-abs-path option.
20967 * common.opt (fprofile-abs-path): New option.
20968 * gcov-io.h (gcov_write_filename): Declare.
20969 * gcov-io.c (gcov_write_filename): New function.
20970 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20971 * profile.c (output_location): Likewise.
20973 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20975 * shring-wrap.c: Revert accidental commit.
20977 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20979 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20981 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20983 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20985 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20986 profile in callgraph edge.
20987 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20988 (apply_scale): Likewise.
20989 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20990 Also scale profile when inlining function with zero profile.
20991 (initialize_cfun): Update exit block profile even when it is zero.
20992 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20993 when profile is read.
20995 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20997 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20998 (CLONE_*): New constants to define the processors we can generate
20999 code for with the target_clone attribute.
21000 (rs6000_clone_map): New array to identify which clone processors
21001 the current program is running on.
21002 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21003 target_clone attribute.
21004 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21005 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21006 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21007 (cpu_expand_builtin): Add support for target_clone attribute.
21008 (rs6000_valid_attribute_p): Allow "default" attribute.
21009 (get_decl_name): New debug function to simplify printing the
21010 current function name in debugging statements.
21011 (rs6000_clone_priority): New functions to support the target_clone
21012 attribute, and be able to generate code to switch between ISA 2.05
21013 through ISA 3.0 (power6 through power9).
21014 (rs6000_compare_version_priority): Likewise.
21015 (rs6000_get_function_versions_dispatcher): Likewise.
21016 (make_resolver_func): Likewise.
21017 (add_condition_to_bb): Likewise.
21018 (dispatch_function_versions): Likewise.
21019 (rs6000_generate_version_dispatcher_body): Likewise.
21020 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21021 (fusion_gpr_load_p): Fix a spacing issue.
21022 * doc/extend.texi (Common Function Attributes): Document that the
21023 PowerPC supports the target_clone attribute.
21025 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21027 * config/arm/arm.h: explain F symbol found in description of ARM
21028 register allocation in its legend.
21030 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21032 * config/mips/frame-header-opt.c: Include profile-count.h.
21033 * config/riscv/riscv.c: Include profile-count.h
21035 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21037 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21039 (sm_set_flag_if_changed): Add bbs field.
21040 (execute_sm_if_changed_flag_set): Pass BBS.
21041 (execute_sm): Update.
21043 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21045 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21048 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21050 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21051 (peephole2): New peephole2 to emit the above.
21052 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21054 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21056 * config/aarch64/aarch64.c (define_peephole2 above
21057 *sub_<shift>_<mode>): New peephole.
21059 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21061 * config/i386/i386.c (make_resolver_func): Update.
21062 * Makefile.in: Add profile-count.h and profile-count.o
21063 * auto-profile.c (afdo_indirect_call): Update to new API.
21064 (afdo_set_bb_count): Update.
21065 (afdo_propagate_edge): Update.
21066 (afdo_propagate_circuit): Update.
21067 (afdo_calculate_branch_prob): Update.
21068 (afdo_annotate_cfg): Update.
21069 * basic-block.h: Include profile-count.h
21070 (struct edge_def): Turn count to profile_count.
21071 (struct basic_block_def): Likewie.
21072 (REG_BR_PROB_BASE): Move to profile-count.h
21073 (RDIV): Move to profile-count.h
21074 * bb-reorder.c (max_entry_count): Turn to profile_count.
21075 (find_traces): Update.
21076 (rotate_loop):Update.
21077 (connect_traces):Update.
21078 (sanitize_hot_paths):Update.
21079 * bt-load.c (migrate_btr_defs): Update.
21080 * cfg.c (RDIV): Remove.
21081 (init_flow): Use alloc_block.
21082 (alloc_block): Uninitialize count.
21083 (unchecked_make_edge): Uninitialize count.
21084 (check_bb_profile): Update.
21085 (dump_edge_info): Update.
21086 (dump_bb_info): Update.
21087 (update_bb_profile_for_threading): Update.
21088 (scale_bbs_frequencies_int): Update.
21089 (scale_bbs_frequencies_gcov_type): Update.
21090 (scale_bbs_frequencies_profile_count): New.
21091 * cfg.h (update_bb_profile_for_threading): Update.
21092 (scale_bbs_frequencies_profile_count): Declare.
21093 * cfgbuild.c (compute_outgoing_frequencies): Update.
21094 (find_many_sub_basic_blocks): Update.
21095 * cfgcleanup.c (try_forward_edges): Update.
21096 (try_crossjump_to_edge): Update.
21097 * cfgexpand.c (expand_gimple_tailcall): Update.
21098 (construct_exit_block): Update.
21099 * cfghooks.c (verify_flow_info): Update.
21100 (dump_bb_for_graph): Update.
21101 (split_edge): Update.
21102 (make_forwarder_block): Update.
21103 (duplicate_block): Update.
21104 (account_profile_record): Update.
21105 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21106 (get_estimated_loop_iterations): Update.
21107 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21108 (single_likely_exit): Update.
21109 * cfgloopmanip.c (scale_loop_profile): Update.
21111 (set_zero_probability): Update.
21112 (lv_adjust_loop_entry_edge): Update.
21113 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21114 (purge_dead_edges): Update.
21115 (rtl_account_profile_record): Update.
21116 * cgraph.c (cgraph_node::create): Uninitialize count.
21117 (symbol_table::create_edge): Uninitialize count.
21118 (cgraph_update_edges_for_call_stmt_node): Update.
21119 (cgraph_edge::dump_edge_flags): Update.
21120 (cgraph_node::dump): Update.
21121 (cgraph_edge::maybe_hot_p): Update.
21122 * cgraph.h: Include profile-count.h
21123 (create_clone), create_edge, create_indirect_edge): Update.
21124 (cgraph_node): Turn count to profile_count.
21125 (cgraph_edge0: Likewise.
21126 (make_speculative, clone): Update.
21127 (create_edge): Update.
21128 (init_lowered_empty_function): Update.
21129 * cgraphclones.c (cgraph_edge::clone): Update.
21130 (duplicate_thunk_for_node): Update.
21131 (cgraph_node::create_clone): Update.
21132 * cgraphunit.c (cgraph_node::analyze): Update.
21133 (cgraph_node::expand_thunk): Update.
21134 * final.c (dump_basic_block_info): Update.
21135 * gimple-streamer-in.c (input_bb): Update.
21136 * gimple-streamer-out.c (output_bb): Update.
21137 * graphite.c (print_global_statistics): Update.
21138 (print_graphite_scop_statistics): Update.
21139 * hsa-brig.c: Include basic-block.h.
21140 * hsa-dump.c: Include basic-block.h.
21141 * hsa-gen.c (T sum_slice): Update.
21142 (convert_switch_statements):Update.
21143 * hsa-regalloc.c: Include basic-block.h.
21144 * ipa-chkp.c (chkp_produce_thunks): Update.
21145 * ipa-cp.c (struct caller_statistics): Update.
21146 (init_caller_stats): Update.
21147 (gather_caller_stats): Update.
21148 (ipcp_cloning_candidate_p): Update.
21149 (good_cloning_opportunity_p): Update.
21150 (get_info_about_necessary_edges): Update.
21151 (dump_profile_updates): Update.
21152 (update_profiling_info): Update.
21153 (update_specialized_profile): Update.
21154 (perhaps_add_new_callers): Update.
21155 (decide_about_value): Update.
21156 (ipa_cp_c_finalize): Update.
21157 * ipa-devirt.c (struct odr_type_warn_count): Update.
21158 (struct decl_warn_count): Update.
21159 (struct final_warning_record): Update.
21160 (possible_polymorphic_call_targets): Update.
21161 (ipa_devirt): Update.
21162 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21163 * ipa-icf.c (sem_function::merge): Update.
21164 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21165 * ipa-inline.c (compute_uninlined_call_time): Update.
21166 (compute_inlined_call_time): Update.
21167 (want_inline_small_function_p): Update.
21168 (want_inline_self_recursive_call_p): Update.
21169 (edge_badness): Update.
21170 (lookup_recursive_calls): Update.
21171 (recursive_inlining): Update.
21172 (inline_small_functions): Update.
21173 (dump_overall_stats): Update.
21174 (dump_inline_stats): Update.
21175 * ipa-profile.c (ipa_profile_generate_summary): Update.
21176 (ipa_propagate_frequency): Update.
21177 (ipa_profile): Update.
21178 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21179 * ipa-utils.c (ipa_merge_profiles): Update.
21180 * loop-doloop.c (doloop_modify): Update.
21181 * loop-unroll.c (report_unroll): Update.
21182 (unroll_loop_runtime_iterations): Update.
21183 * lto-cgraph.c (lto_output_edge): Update.
21184 (lto_output_node): Update.
21185 (input_node): Update.
21186 (input_edge): Update.
21187 (merge_profile_summaries): Update.
21188 * lto-streamer-in.c (input_cfg): Update.
21189 * lto-streamer-out.c (output_cfg): Update.
21190 * mcf.c (create_fixup_graph): Update.
21191 (adjust_cfg_counts): Update.
21192 (sum_edge_counts): Update.
21193 * modulo-sched.c (sms_schedule): Update.
21194 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21195 * predict.c (maybe_hot_count_p): Update.
21196 (probably_never_executed): Update.
21197 (dump_prediction): Update.
21198 (combine_predictions_for_bb): Update.
21199 (propagate_freq): Update.
21200 (handle_missing_profiles): Update.
21201 (counts_to_freqs): Update.
21202 (rebuild_frequencies): Update.
21203 (force_edge_cold): Update.
21204 * predict.h: Include profile-count.h
21205 (maybe_hot_count_p, counts_to_freqs): UPdate.
21206 * print-rtl-function.c: Do not include cfg.h
21207 * print-rtl.c: Include basic-block.h
21208 * profile-count.c: New file.
21209 * profile-count.h: New file.
21210 * profile.c (is_edge_inconsistent): Update.
21211 (correct_negative_edge_counts): Update.
21212 (is_inconsistent): Update.
21213 (set_bb_counts): Update.
21214 (read_profile_edge_counts): Update.
21215 (compute_frequency_overlap): Update.
21216 (compute_branch_probabilities): Update; Initialize and deinitialize
21218 (branch_prob): Update.
21219 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21220 (edge_gcov_count): New.
21221 (bb_gcov_count): New.
21222 * shrink-wrap.c (try_shrink_wrapping): Update.
21223 * tracer.c (better_p): Update.
21224 * trans-mem.c (expand_transaction): Update.
21225 (ipa_tm_insert_irr_call): Update.
21226 (ipa_tm_insert_gettmclone_call): Update.
21227 * tree-call-cdce.c: Update.
21228 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21229 (gimple_duplicate_sese_tail): Update.
21230 (gimple_account_profile_record): Update.
21231 (execute_fixup_cfg): Update.
21232 * tree-inline.c (copy_bb): Update.
21233 (copy_edges_for_bb): Update.
21234 (initialize_cfun): Update.
21235 (freqs_to_counts): Update.
21236 (copy_cfg_body): Update.
21237 (expand_call_inline): Update.
21238 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21239 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21240 (try_unroll_loop_completely): Update.
21241 (try_peel_loop): Update.
21242 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21243 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21244 * tree-ssa-loop-split.c (connect_loops): Update.
21245 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21246 * tree-ssa-reassoc.c (branch_fixup): Update.
21247 * tree-ssa-tail-merge.c (replace_block_by): Update.
21248 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21249 (compute_path_counts): Update.
21250 (update_profile): Update.
21251 (recompute_probabilities): Update.
21252 (update_joiner_offpath_counts): Update.
21253 (estimated_freqs_path): Update.
21254 (freqs_to_counts_path): Update.
21255 (clear_counts_path): Update.
21256 (ssa_fix_duplicate_block_edges): Update.
21257 (duplicate_thread_path): Update.
21258 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21259 (struct switch_conv_info): Update.
21260 * tree-tailcall.c (decrease_profile): Update.
21261 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21262 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21263 * value-prof.c (check_counter): Update.
21264 (gimple_divmod_fixed_value): Update.
21265 (gimple_mod_pow2): Update.
21266 (gimple_mod_subtract): Update.
21267 (gimple_ic_transform): Update.
21268 (gimple_stringop_fixed_value): Update.
21269 * value-prof.h (gimple_ic): Update.
21271 2017-06-02 Carl Love <cel@us.ibm.com>
21273 * config/rs6000/rs6000-c: Add support for built-in functions
21274 vector double vec_doublee (vector signed int);
21275 vector double vec_doublee (vector unsigned int);
21276 vector double vec_doublee (vector float);
21277 vector double vec_doubleh (vector signed int);
21278 vector double vec_doubleh (vector unsigned int);
21279 vector double vec_doubleh (vector float);
21280 vector double vec_doublel (vector signed int);
21281 vector double vec_doublel (vector unsigned int);
21282 vector double vec_doublel (vector float);
21283 vector double vec_doubleo (vector signed int);
21284 vector double vec_doubleo (vector unsigned int);
21285 vector double vec_doubleo (vector float);.
21286 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21287 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21289 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21290 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21291 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21293 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21294 vec_doublel, vec_doubleh.
21295 * doc/extend.texi: Update the built-in documentation file for the
21296 new built-in functions.
21298 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21301 * ipa-inline-analysis.c (free_growth_caches): Set
21302 edge_removal_hook_holder to NULL after removing it.
21304 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21306 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21307 comparision with zero.
21309 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21310 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21311 for early expansion of vec_min and vec_max builtins.
21312 (builtin_function_type): Add min/max unsigned variants to those
21313 identified as having unsigned arguments.
21315 2017-06-02 Olivier Hainque <hainque@adacore.com>
21317 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21319 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21321 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21322 Use VALL_F16 iterator rather than VALL.
21324 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21326 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21327 Emit CBNZ inside loop when doing a strong exchange and comparing
21328 against zero. Generate the CC flags after the loop.
21330 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21332 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21333 (dl_section_ref): New.
21334 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21335 On AIX, append an expression to subtract the size of the
21336 section length to dl_section_ref.
21338 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21340 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21341 for early expansion of vector absolute builtins.
21343 2017-06-02 Richard Biener <rguenther@suse.de>
21345 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21346 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21348 2017-06-02 Richard Biener <rguenther@suse.de>
21350 PR tree-optimization/80948
21351 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21352 stmt order as well.
21354 2017-06-02 Richard Biener <rguenther@suse.de>
21356 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21358 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21359 for inductions as relevant.
21361 2017-06-02 Richard Biener <rguenther@suse.de>
21363 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21364 (vectorizable_induction): ... this. Remove dead code.
21366 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21368 * builtins. (expand_builtin_alloca): Remove second parameter and
21369 infer its value from the first parameter instead.
21370 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21372 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21374 PR rtl-optimization/80903
21375 * loop-doloop.c (add_test): Unshare sequence.
21377 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21379 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21381 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21383 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21385 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21386 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21387 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21388 (xlogue_layout::get_stub_rtx): Make static.
21389 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21390 (xlogue_layout::compute_stub_managed_regs): Rename to...
21391 (xlogue_layout::count_stub_managed_regs): ...this.
21392 (xlogue_layout::is_stub_managed_reg): New function.
21393 (xlogue_layout::m_stub_names): Rename to...
21394 (xlogue_layout::s_stub_names): ...this, make static.
21395 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21396 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21397 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21398 xlogue_layout::s_stub_names): Instantiate statics.
21399 (stub_managed_regs): Remove.
21400 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21401 (disable_call_ms2sysv_xlogues): Rename to...
21402 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21403 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21405 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21406 change after reload_completed.
21407 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21409 (ix86_expand_prologue): Likewise.
21410 (ix86_expand_epilogue): Likewise.
21411 (ix86_expand_split_stack_prologue): Likewise.
21412 (ix86_compute_frame_layout): Remove frame parameter ...
21413 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21414 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21416 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21417 (ix86_frame): Move from here ...
21418 * config/i386/i386.h (ix86_frame): ... to here.
21419 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21420 complete ix86_frame data structure instead. Remove some_ld_name.
21422 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21424 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21425 symbols that hold a DECL_VALUE_EXPR.
21427 2017-06-01 Martin Jambor <mjambor@suse.cz>
21429 PR tree-optimization/80898
21430 * tree-sra.c (process_subtree_disqualification): Removed.
21431 (disqualify_candidate): Do not acll
21432 process_subtree_disqualification.
21433 (subtree_mark_written_and_enqueue): New function.
21434 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21435 RHS has been disqualified and re-queue LHS if necessary. Apart
21436 from that, ignore disqualified RHS.
21438 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21440 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21441 address fetch for z10 or later.
21443 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21445 * config/arc/arc.md (tst_movb): Add guard when splitting.
21447 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21449 * config/arc/arc.c (arc_can_eliminate): Test against
21450 arc_frame_pointer_needed.
21452 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21454 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21455 to prevent store reordering.
21456 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21457 (type): Add block type.
21458 (stack_tie): Define special instruction to be used in
21461 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21463 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21464 constraint. It is not valid for the pattern.
21465 (noncommutative_binary_comparison): Likewise.
21467 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21469 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21472 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21474 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21475 be used by the reg-alloc.
21477 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21479 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21480 reg-alloc when having mul64 or mul32x16 instructions.
21481 (mulsidi3): Likewise.
21482 (umulsidi3): Likewise.
21483 (mulsi32x16): New pattern.
21484 (mulsi64): Likewise.
21485 (mulsidi64): Likewise.
21486 (umulsidi64): Likewise.
21487 (MUL32x16_REG): Define.
21488 (mul64_600): Use MUL32x16_REG.
21489 (mac64_600): Likewise.
21490 (umul64_600): Likewise.
21491 (umac64_600): Likewise.
21493 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21495 * config/arc/arc.md (mulsi3_700): Make it commutative.
21497 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21499 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21501 (*sign_extendsidi2_insn): Likewise for movstosw.
21503 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21505 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21506 the type of the input discriminant value. Convert the
21507 discriminant value of signedness vary.
21509 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21511 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21512 Add to -Wall section.
21514 2017-06-01 Richard Biener <rguenther@suse.de>
21516 PR middle-end/66313
21517 * fold-const.c (fold_plusminus_mult_expr): If the factored
21518 factor may be zero use a wrapping type for the inner operation.
21519 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21520 and handle moved defs.
21521 (process_assignment): Properly guard the unary op case. Return a
21522 tri-state indicating that moving the stmt before the call may allow
21523 to continue. Pass through to_move.
21524 (find_tail_calls): Handle moving unrelated defs before
21527 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21530 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21531 splitter result in the canonical way.
21533 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21535 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21536 also for 32bit target. Update insn attributes.
21537 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21539 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21541 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21542 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21543 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21544 (_mm_maskz_min_ss): New intrinsics.
21546 2017-05-31 Martin Liska <mliska@suse.cz>
21548 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21549 Change comment style to one we normally use.
21550 (vectorizable_reduction): Likewise.
21551 (vectorizable_induction): Likewise.
21552 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21553 (vectorizable_call): Likewise.
21554 (vectorizable_simd_clone_call): Likewise.
21555 (vectorizable_conversion): Likewise.
21556 (vectorizable_assignment): Likewise.
21557 (vectorizable_shift): Likewise.
21558 (vectorizable_operation): Likewise.
21559 (vectorizable_store): Likewise.
21560 (vectorizable_load): Likewise.
21561 * tree-vectorizer.h: Likewise.
21563 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21565 * passes.c (emergency_dump_function): New.
21566 * tree-pass.h (emergency_dump_function): Declare.
21567 * plugin.c (plugins_internal_error_function): Remove.
21568 * plugin.h (plugins_internal_error_function): Remove declaration.
21569 * toplev.c (internal_error_function): New static function. Use it...
21570 (general_init): ...here.
21572 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21574 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21575 (arc_rtx_costs): Add costs for new patterns.
21576 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21577 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21579 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21581 * tree-ssa-strlen.c (get_next_strinfo): New function.
21582 (get_stridx_plus_constant): Use it.
21583 (zero_length_string): Likewise.
21584 (adjust_related_strinfos): Likewise.
21585 (adjust_last_stmt): Likewise.
21587 2017-05-31 Richard Biener <rguenther@suse.de>
21590 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21591 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21593 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21595 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21596 loop_vinfo argument and use of dependence distance vectors.
21597 Check instead whether the two references differ only in their
21598 initial value and assume that they have the same alignment if the
21599 difference is a multiple of the vector alignment.
21600 (vect_analyze_data_refs_alignment): Update call accordingly.
21602 2017-05-31 Martin Liska <mliska@suse.cz>
21605 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21607 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21609 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21610 (create_intersect_range_checks): Move from ...
21611 * tree-data-ref.c (create_intersect_range_checks_index)
21612 (create_intersect_range_checks): ... to here.
21613 (create_runtime_alias_checks): New function factored from ...
21614 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21615 here. Call above function.
21616 * tree-data-ref.h (create_runtime_alias_checks): New function.
21618 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21620 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21621 segment length for dr_b and compute it in wide_int.
21623 2017-05-31 Richard Biener <rguenther@suse.de>
21625 PR tree-optimization/80906
21626 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21627 and pass through iv_map.
21628 (copy_bb_and_scalar_dependences): Adjust.
21629 (translate_pending_phi_nodes): Likewise.
21630 (copy_loop_close_phi_args): Handle code-generating IVs instead
21633 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21635 * diagnostic-color.c (color_dict): Add "type-diff".
21636 (parse_gcc_colors): Update comment.
21637 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21638 -fdiagnostics-show-template-tree and -fno-elide-type.
21639 (GCC_COLORS): Add type-diff to example.
21641 (-fdiagnostics-show-template-tree): New.
21642 (-fno-elide-type): New.
21643 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21644 the pp_format_decoder callback. Call any m_format_postprocessor's
21646 (pretty_printer::pretty_printer): Initialize
21647 m_format_postprocessor.
21648 (pretty_printer::~pretty_printer): Delete any
21649 m_format_postprocessor.
21650 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21651 (class format_postprocessor): New class.
21652 (struct pretty_printer::format_decoder): Document the new parameters.
21653 (struct pretty_printer::m_format_postprocessor): New field.
21654 * tree-diagnostic.c (default_tree_printer): Update for new
21655 bool and const char ** params.
21656 * tree-diagnostic.h (default_tree_printer): Likewise.
21658 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21660 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21661 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21662 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21663 rs6000_gen_cell_microcode code.
21664 (rs6000_final_prescan_insn): Delete.
21665 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21666 "warn-cell-microcode" entries.
21667 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21668 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21669 throughout. Change cc_reg_not_micro_cr0_operand to
21670 cc_reg_not_cr0_operand throughout.
21671 (*extendhi<mode>2_noload): Delete.
21672 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21673 (mwarn-cell-microcode): Delete.
21674 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21675 -mgen-cell-microcode and -mwarn-cell-microcode.
21677 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21680 * config/i386/constraints.md (Yd): New constraint.
21682 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21683 and (?Yd, r) alternatives. Update insn attributes.
21684 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21685 and (?*Yd, r) alternatives. Update insn attributes.
21686 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21688 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21690 * gimplify.c (gimplify_modify_expr): Don't create a
21691 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21694 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21696 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21698 2017-05-30 Richard Biener <rguenther@suse.de>
21700 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21701 and reduc_def fields.
21702 (STMT_VINFO_REDUC_TYPE): New define.
21703 (STMT_VINFO_REDUC_DEF): Likewise.
21704 (vect_force_simple_reduction): Adjust prototype.
21705 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21706 (vect_is_simple_reduction): Remove check_reduction argument.
21707 (vect_force_simple_reduction): Adjust and set
21708 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21709 (vectorizable_reduction): Do not re-do reduction analysis
21710 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21711 * tree-parloops.c (gather_scalar_reductions): Adjust.
21713 2017-05-30 Richard Biener <rguenther@suse.de>
21715 PR middle-end/80901
21716 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21719 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21721 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21722 Introduce unknown_misalignment parameter and remove vf.
21723 (vect_peeling_hash_get_lowest_cost):
21724 Pass unknown_misalignment parameter.
21725 (vect_enhance_data_refs_alignment):
21726 Fix unsupportable data ref treatment.
21728 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21730 * tree-vect-data-refs.c (vect_get_data_access_cost):
21731 Workaround for SLP handling.
21732 (vect_enhance_data_refs_alignment):
21733 Compute costs for doing no peeling at all, compare to the best
21734 peeling costs so far and avoid peeling if cheaper.
21736 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21738 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21739 Return peeling info and set costs to zero for unlimited cost
21741 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21742 with unknown misalignment. Compute and costs for unknown
21743 misalignment, compare them to the costs for known misalignment
21744 and choose the cheapest for peeling.
21746 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21748 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21749 (vect_get_peeling_costs_all_drs): Create function.
21750 (vect_peeling_hash_get_lowest_cost):
21751 Use vect_get_peeling_costs_all_drs.
21752 (vect_peeling_supportable): Create function.
21753 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21755 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21757 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21758 DR_HAS_NEGATIVE_STEP.
21759 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21760 (vect_enhance_data_refs_alignment): Use.
21761 (vect_duplicate_ssa_name_ptr_info): Use.
21762 * tree-vectorizer.h (dr_misalignment): Use.
21763 (known_alignment_for_access_p): Use.
21765 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21768 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21770 (has_section_name): New function.
21772 2017-05-30 Martin Liska <mliska@suse.cz>
21775 * auto-profile.c (get_function_decl_from_block): Fix
21778 2017-05-30 Richard Biener <rguenther@suse.de>
21780 PR middle-end/80876
21781 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21783 2017-05-30 Martin Liska <mliska@suse.cz>
21785 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21786 * dumpfile.h (struct dump_file_info): Remove ctors.
21788 2017-05-30 Martin Liska <mliska@suse.cz>
21790 * predict.def: Fix GNU coding style.
21792 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21794 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21795 Mark 'to' argument with ATTRIBUTE_UNUSED.
21797 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21799 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21800 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21801 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21804 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21806 * doc/install.texi (Options specification): Restore entry of
21807 --enable-sjlj-exceptions.
21809 2017-05-27 Michael Eager <eager@eagercon.com>
21812 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21814 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21816 * config/microblaze/microblaze.h
21817 (FIXED_REGISTERS): Update in macro.
21818 (CALL_USED_REGISTERS): Update in macro.
21820 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21822 * doc/install.texi: Add links to macOS binary distributions.
21824 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21828 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21830 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21832 2017-05-26 Martin Liska <mliska@suse.cz>
21834 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21836 2017-05-26 Martin Liska <mliska@suse.cz>
21838 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21839 always leading ';; '.
21840 (dump_bb_info): Likewise.
21841 (brief_dump_cfg): Likewise.
21842 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21843 * dumpfile.c: Remove usage of TDF_VERBOSE.
21844 * dumpfile.h (enum dump_kind): Likewise.
21845 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21846 * print-tree.c (debug_verbose): Remove.
21847 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21848 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21849 * tree-diagnostic.c (default_tree_printer): Replace
21850 TDF_DIAGNOSTIC with TDF_SLIM.
21852 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21854 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21855 in parameter loop, rather than loop_vinfo.
21856 (create_intersect_range_checks): Ditto.
21857 (vect_create_cond_for_alias_checks): Update call to above functions.
21859 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21861 PR tree-optimization/80815
21862 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21863 for merging runtime alias checks. Handle negative DR_STEPs.
21865 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21867 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21869 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21870 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21871 out code pruning runtime alias checks.
21872 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21873 factored out from above.
21874 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21876 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21878 (prune_runtime_alias_test_list): New decalaration.
21880 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21882 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21883 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21884 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21885 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21886 (operator==, comp_dr_with_seg_len_pair): Ditto.
21887 (vect_prune_runtime_alias_test_list): Ditto.
21889 2017-05-26 Martin Liska <mliska@suse.cz>
21892 * params.def: Bound partial-inlining-entry-probability param.
21894 2017-05-26 Marek Polacek <polacek@redhat.com>
21897 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21900 2017-05-26 Richard Biener <rguenther@suse.de>
21902 PR tree-optimization/80842
21903 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21906 2017-05-26 Richard Biener <rguenther@suse.de>
21908 PR tree-optimization/80844
21909 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21911 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21913 * doc/md.texi (Machine Constraints): Update x86 family
21914 machine constraints section to match 'config/i386/constraints.md'.
21916 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21918 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21920 2017-05-25 Nathan Sidwell <nathan@acm.org>
21922 * doc/invoke.texi (--enable-languages): Update documentation.
21924 2017-05-25 Martin Liska <mliska@suse.cz>
21926 * dumpfile.c: Add TDF_FOLDING.
21927 * dumpfile.h (enum dump_kind): Likewise.
21928 * genmatch.c (dt_simplify::gen_1): Use it.
21930 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21932 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21934 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21936 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21937 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21939 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21941 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21942 * match.pd (X == C): Rewrite it here.
21943 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21944 with_certain_nonzero_bits2): New predicates.
21945 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21947 2017-05-24 Nathan Sidwell <nathan@acm.org>
21949 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21952 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21955 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21957 * config/powerpcspe: New port. Files are copied from the rs6000
21958 port, with "rs6000" in filenames replaced by "powerpcspe".
21960 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21962 PR rtl-optimization/80754
21963 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21965 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21967 * config/sparc/sparc.md (length): Return the correct value for -mflat
21968 sibcalls to match output_sibcall.
21970 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21974 * config/rs6000/rs6000.c (struct machine_function): Add new field
21976 (rs6000_get_separate_components): Init that field, use it.
21977 (rs6000_components_for_bb): Use the field.
21979 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21981 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21983 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21985 PR middle-end/80823
21986 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21988 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21991 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21992 address against address_operand predicate.
21993 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21995 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21997 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21998 parameters passed indirectly.
22000 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22002 * config/i386/i386.md (*movdi_internal): Remove SSE4
22003 alternative 18 (?r, *v). Update insn attributes.
22004 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22005 Update insn attributes.
22006 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22007 Update insn attributes.
22008 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22009 alternative 1 (r, v). Remove isa attribute.
22010 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22011 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22012 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22014 2017-05-23 Tom de Vries <tom@codesourcery.com>
22016 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22019 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22021 * cgraphunit.c (symbol_table::process_new_functions): Update.
22022 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22023 (inline_generate_summary): Rename to ...
22024 (ipa_fn_summary_generate): ... this one.
22025 (inline_read_summary): Rename to ...
22026 (ipa_fn_summary_read): ... this one.
22027 (inline_write_summary): Rename to ...
22028 (ipa_fn_summary_write): ... this one.
22029 (inline_free_summary): Rename to ...
22030 (ipa_free_fn_summary): ... this one.
22031 (pass_data_local_fn_summary, pass_local_fn_summary,
22032 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22033 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22034 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22035 make_pass_ipa_fn_summary): New.
22036 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22037 inline_write_summary, inline_free_summary): Remove.
22038 (ipa_free_fn_summary) : New.
22039 * ipa-inline.c (ipa_inline): Update.
22040 (pass_ipa_inline): Do not generate summaries.
22041 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22043 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22044 and add pass_ipa_fn_summary.
22045 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22047 (make_pass_inline_parameters): Remove.
22049 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22051 * omp-low.c (struct omp_context): Remove "default_kind" member.
22054 * omp-offload.c (execute_oacc_device_lower): Remove the
22055 parallelism dimensions function attributes for unparallelized
22056 OpenACC kernels constructs.
22058 2017-05-23 Martin Liska <mliska@suse.cz>
22060 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22062 (cgraph_edge::make_speculative): Likewise.
22063 (cgraph_edge::resolve_speculation): Likewise.
22064 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22065 (cgraph_node::dump): Likewise.
22066 * cgraph.h: Likewise.
22067 * cgraphunit.c (analyze_functions): Likewise.
22068 (symbol_table::compile): Likewise.
22069 * ipa-cp.c (print_all_lattices): Likewise.
22070 (determine_versionability): Likewise.
22071 (initialize_node_lattices): Likewise.
22072 (ipcp_verify_propagated_values): Likewise.
22073 (estimate_local_effects): Likewise.
22074 (update_profiling_info): Likewise.
22075 (create_specialized_node): Likewise.
22076 (perhaps_add_new_callers): Likewise.
22077 (decide_about_value): Likewise.
22078 (decide_whether_version_node): Likewise.
22079 (identify_dead_nodes): Likewise.
22080 (ipcp_store_bits_results): Likewise.
22081 * ipa-devirt.c (dump_targets): Likewise.
22082 (ipa_devirt): Likewise.
22083 * ipa-icf.c (sem_item::dump): Likewise.
22084 (sem_function::equals): Likewise.
22085 (sem_variable::equals): Likewise.
22086 (sem_item_optimizer::read_section): Likewise.
22087 (sem_item_optimizer::execute): Likewise.
22088 (congruence_class::dump): Likewise.
22089 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22090 (dump_inline_summary): Likewise.
22091 (estimate_node_size_and_time): Likewise.
22092 (inline_analyze_function): Likewise.
22093 * ipa-inline-transform.c (inline_call): Likewise.
22094 * ipa-inline.c (report_inline_failed_reason): Likewise.
22095 (want_early_inline_function_p): Likewise.
22096 (edge_badness): Likewise.
22097 (update_edge_key): Likewise.
22098 (inline_small_functions): Likewise.
22099 * ipa-profile.c (ipa_profile): Likewise.
22100 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22101 (ipa_make_edge_direct_to_target): Likewise.
22102 (remove_described_reference): Likewise.
22103 (ipa_impossible_devirt_target): Likewise.
22104 (propagate_controlled_uses): Likewise.
22105 (ipa_print_node_params): Likewise.
22106 (ipcp_transform_function): Likewise.
22107 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22108 (propagate_pure_const): Likewise.
22109 * ipa-reference.c (generate_summary): Likewise.
22110 (read_write_all_from_decl): Likewise.
22111 (propagate): Likewise.
22112 (ipa_reference_read_optimization_summary): Likewise.
22113 * ipa-utils.c (ipa_merge_profiles): Likewise.
22114 * ipa.c (walk_polymorphic_call_targets): Likewise.
22115 (symbol_table::remove_unreachable_nodes): Likewise.
22116 (ipa_single_use): Likewise.
22117 * passes.c (execute_todo): Likewise.
22118 * predict.c (drop_profile): Likewise.
22119 * symtab.c (symtab_node::get_dump_name): New function.
22120 (symtab_node::dump_name): Likewise.
22121 (symtab_node::dump_asm_name): Likewise.
22122 (symtab_node::dump_references): Likewise.
22123 (symtab_node::dump_referring): Likewise.
22124 (symtab_node::dump_base): Likewise.
22125 (symtab_node::debug_symtab): Likewise.
22126 * tree-sra.c (convert_callers_for_node): Likewise.
22127 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22128 * value-prof.c (init_node_map): Likewise.
22130 2017-05-23 Martin Liska <mliska@suse.cz>
22132 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22133 and symtab_node::debug_symtab to symbol_table::debug.
22134 * cgraphunit.c (analyze_functions): Use the renamed function.
22135 (symbol_table::compile): Likewise.
22136 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22137 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22138 * passes.c (execute_todo): Likewise.
22139 * symtab.c (symbol_table::dump): New function.
22140 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22142 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22144 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22145 that nonconst implies exec.
22147 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22149 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22150 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22151 (inline_edge_summary_vec): Turn into ...
22152 (ipa_call_summaries): ... this one.
22153 (redirect_to_unreachable, edge_set_predicate,
22154 evaluate_properties_for_edge, inline_summary_alloc,
22155 reset_ipa_call_summary, reset_inline_summary,
22156 inline_summary_t::duplicate): Update.
22157 (inline_edge_duplication_hook): Turn to ...
22158 (ipa_call_summary_t::duplicate): ... this one.
22159 (inline_edge_removal_hook): Turn to ...
22160 (ipa_call_summary_t::remove): ... this one.
22161 (dump_inline_edge_summary): Turn to ...
22162 (dump_ipa_call_summary): ... this one.
22163 (estimate_function_body_sizes): Update.
22164 (inline_update_callee_summaries): Update.
22165 (remap_edge_change_prob): Update.
22166 (remap_edge_summaries): Update.
22167 (inline_merge_summary): Update.
22168 (do_estimate_edge_time): Update.
22169 (inline_generate_summary): Update.
22170 (inline_read_section): Update.
22171 (inline_read_summary): Update.
22172 (inline_free_summary): Update.
22173 * ipa-inline.c (can_inline_edge_p): Update.
22174 (compute_inlined_call_time): Update.
22175 (want_inline_small_function_p): Update.
22176 (edge_badness): Update.
22177 (early_inliner): Update.
22178 * ipa-inline.h (inline_edge_summary): Turn to ...
22179 (ipa_call_summary): ... this one.
22180 (ipa_call_summary_t): New class.
22181 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22182 (ipa_call_summaries): New.
22183 (inline_edge_summary): Remove.
22184 (estimate_edge_growth): Update.
22185 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22186 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22187 * ipa-split.c (execute_split_functions): Update.
22188 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22190 2017-05-23 Tom de Vries <tom@codesourcery.com>
22192 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22193 attributes): Document rdrand effective target.
22195 2017-05-23 Tom de Vries <tom@codesourcery.com>
22197 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22198 attributes): Sort alphabetically.
22200 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22202 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22204 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22207 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22208 V2DF/V2DI splat into two separate patterns, one that handles
22209 registers, and the other that only handles memory. Drop support
22210 for splatting from a GPR on ISA 2.07 and then splitting the
22211 splat into direct move and splat.
22212 (vsx_splat_<mode>_reg): Likewise.
22213 (vsx_splat_<mode>_mem): Likewise.
22215 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22217 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22219 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22221 PR middle-end/80809
22222 * omp-low.c (finish_taskreg_remap): New function.
22223 (finish_taskreg_scan): If unit size of ctx->record_type
22224 is non-constant, unshare the size expression and replace
22225 decls in it with possible outer var refs.
22227 PR middle-end/80809
22228 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22229 GOVD_SHARED rather than GOVD_PRIVATE with it.
22230 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22231 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22233 PR middle-end/80853
22234 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22235 as last argument to build_outer_var_ref for pointer bases of array
22236 section reductions.
22238 2017-05-19 Martin Sebor <msebor@redhat.com>
22240 * print-tree.c (print_node): Print DECL_READ_P flag.
22242 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22244 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22245 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22246 * cgraph.c: Likewise.
22247 * cgraphunit.c: Likewise.
22248 * gengtype.c: Likewise.
22249 * ipa-cp.c: Likewise.
22250 * ipa-devirt.c: Likewise.
22251 * ipa-icf.c: Likewise.
22252 * ipa-predicate.c: Likewise.
22253 * ipa-profile.c: Likewise.
22254 * ipa-prop.c: Likewise.
22255 * ipa-split.c: Likewise.
22257 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22258 edge_predicate_pool, dump_inline_hints,
22259 inline_summary::account_size_time, redirect_to_unreachable,
22260 edge_set_predicate, set_hint_predicate,
22261 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22262 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22263 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22264 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22265 ipa_call_summary_t::remove, initialize_growth_caches,
22266 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22267 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22268 mark_modified, unmodified_parm_1, unmodified_parm,
22269 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22270 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22271 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22272 will_be_nonconstant_predicate, record_modified_bb_info,
22273 get_minimal_bb, record_modified, param_change_prob,
22274 phi_result_unknown_predicate, predicate_for_phi_result,
22275 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22276 estimate_function_body_sizes, compute_inline_parameters,
22277 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22278 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22279 inline_update_callee_summaries, remap_edge_change_prob,
22280 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22281 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22282 inline_analyze_function, inline_summary_t::insert,
22283 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22284 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22285 inline_free_summary): Move to ipa-fnsummary.h
22286 (predicate_t): Remove.
22287 * ipa-fnsummary.c: New file.
22288 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22289 (enum inline_hints_vals, inline_hints, agg_position_info,
22290 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22291 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22292 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22293 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22294 inline_read_summary, inline_write_summary, inline_free_summary,
22295 inline_analyze_function, initialize_inline_failed,
22296 inline_merge_summary, inline_update_overall_summary,
22297 compute_inline_parameters): Move to ipa-fnsummary.h
22298 * ipa-fnsummary.h: New file.
22299 * ipa-inline-transform.h: Include ipa-inline.h.
22300 * ipa-inline.c: LIkewise.
22302 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22304 * ipa-inline.c (edge_badness): Use inlined_time instead of
22305 inline_summaries->get.
22307 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22309 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22311 2017-05-22 Nathan Sidwell <nathan@acm.org>
22313 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22314 (fdump-lang): Document 'raw' option.
22315 * dumpfile.h (TDI_tu): Delete.
22316 * dumpfile.c (dump_files): Remove translation-unit.
22317 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22319 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22321 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22322 command option from $(AWK) call.
22323 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22324 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22325 [FORMAT]: Remove handling of variable.
22326 * config/avr/t-multilib: Regenerate.
22328 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22330 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22332 (dump_inline_summary): Do not print self_time.
22333 (estimate_function_body_sizes): Do not set self_time.
22334 (compute_inline_parameters): Likewise.
22335 (inline_read_section, inline_write_summary): Do not stream self_time.
22336 * ipa-inline.h (inline_summary): Drop self_time.
22338 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22340 * ipa-inline-analysis.c (account_size_time): Rename to ...
22341 (inline_summary::account_size_time): ... this one.
22342 (reset_ipa_call_summary): Turn to ...
22343 (ipa_call_summary::reset): ... this one.
22344 (reset_inline_summary): Turn to ...
22345 (inline_summary::reset): ... this one.
22346 (inline_summary_t::remove): Update.
22347 (inline_summary_t::duplicate): Update.
22348 (ipa_call_summary_t::remove): Update.
22349 (dump_inline_summary): Update.
22350 (estimate_function_body_sizes): Update.
22351 (compute_inline_parameters): Update.
22352 (estimate_node_size_and_time): Update.
22353 (inline_merge_summary): Update.
22354 (inline_update_overall_summary): Update.
22355 (inline_read_section): Update.
22356 (inline_write_summary): Update.
22357 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22358 add account_size_time and reset member functions.
22359 (ipa_call_summary): Add reset function.
22360 * ipa-predicate.h (predicate::operator &): Constify.
22362 2017-05-22 Richard Biener <rguenther@suse.de>
22364 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22366 2017-05-19 Jason Merrill <jason@redhat.com>
22368 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22370 2017-05-19 Marek Polacek <polacek@redhat.com>
22373 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22374 TYPE_OVERFLOW_WRAPS checks.
22376 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22378 * tree-core.h (enum omp_clause_default_kind): Add
22379 "OMP_CLAUSE_DEFAULT_PRESENT".
22380 * tree-pretty-print.c (dump_omp_clause): Handle it.
22381 * gimplify.c (enum gimplify_omp_var_data): Add
22382 "GOVD_MAP_FORCE_PRESENT".
22383 (gimplify_adjust_omp_clauses_1): Map it to
22384 "GOMP_MAP_FORCE_PRESENT".
22385 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22387 * gimplify.c (oacc_default_clause): Clarify.
22389 2017-05-19 Nathan Sidwell <nathan@acm.org>
22391 LANG_HOOK_REGISTER_DUMPS
22392 * toplev.c (general_init): Call register dump lang hook.
22393 * doc/invoke.texi: Document -fdump-lang option family.
22394 * dumpfile.c (dump_files): Remove class dump here.
22395 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22396 * dumpfile.h (tree_dump_index): Remove TDI_class.
22397 * langhooks-def.h (lhd_register_dumps): Declare.
22398 (LANG_HOOKS_REGISTER_DUMPS): Define.
22399 (LANG_HOOKS_INITIALIZER): Add it.
22400 * langhooks.c (lhd_register_dumps): Define.
22401 * langhooks.h (struct lang_hooks): Add register_dumps.
22403 2017-05-19 Nathan Sidwell <nathan@acm.org>
22405 * context.h (context::set_passes): New.
22406 * context.c (context::context): Do not create pass manager.
22407 * toplev.c (general_init): Create pass manager here.
22409 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22411 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22412 use this splitter if two add or or instructions would also work for
22413 the constant we want to generate.
22415 2017-05-19 Richard Biener <rguenther@suse.de>
22418 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22419 predicate evaluation.
22421 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22423 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22425 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22428 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22430 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22431 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22432 (inline_edge_summary_vec): Turn into ...
22433 (ipa_call_summaries): ... this one.
22434 (redirect_to_unreachable, edge_set_predicate,
22435 evaluate_properties_for_edge, inline_summary_alloc,
22436 reset_ipa_call_summary, reset_inline_summary,
22437 inline_summary_t::duplicate): Update.
22438 (inline_edge_duplication_hook): Turn to ...
22439 (ipa_call_summary_t::duplicate): ... this one.
22440 (inline_edge_removal_hook): Turn to ...
22441 (ipa_call_summary_t::remove): ... this one.
22442 (dump_inline_edge_summary): Turn to ...
22443 (dump_ipa_call_summary): ... this one.
22444 (estimate_function_body_sizes): Update.
22445 (inline_update_callee_summaries): Update.
22446 (remap_edge_change_prob): Update.
22447 (remap_edge_summaries): Update.
22448 (inline_merge_summary): Update.
22449 (do_estimate_edge_time): Update.
22450 (inline_generate_summary): Update.
22451 (inline_read_section): Update.
22452 (inline_read_summary): Update.
22453 (inline_free_summary): Update.
22454 * ipa-inline.c (can_inline_edge_p): Update.
22455 (compute_inlined_call_time): Update.
22456 (want_inline_small_function_p): Update.
22457 (edge_badness): Update.
22458 (early_inliner): Update.
22459 * ipa-inline.h (inline_edge_summary): Turn to ...
22460 (ipa_call_summary): ... this one.
22461 (ipa_call_summary_t): New class.
22462 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22463 (ipa_call_summaries): New.
22464 (inline_edge_summary): Remove.
22465 (estimate_edge_growth): Update.
22466 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22467 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22468 * ipa-split.c (execute_split_functions): Update.
22469 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22471 2017-05-19 Richard Biener <rguenther@suse.de>
22473 PR middle-end/80764
22474 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22476 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22478 * config/rs6000/rs6000.c (struct machine_function): Add field
22479 fpr_is_wrapped_separately.
22480 (rs6000_get_separate_components): Use 64 components. Handle the
22481 new FPR components.
22482 (rs6000_components_for_bb): Handle the FPR components.
22483 (rs6000_emit_prologue_components): Handle the FPR components.
22484 (rs6000_emit_epilogue_components): Handle the FPR components.
22485 (rs6000_set_handled_components): Handle the FPR components.
22486 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22487 that are already separately shrink-wrapped.
22488 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22489 that are already separately shrink-wrapped.
22491 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22494 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22497 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22498 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22499 cases where the register allocator uses a move and an offsettable
22500 memory operation to/from a FPR register on ISA 2.06/2.07.
22501 (define_peephole2 for Altivec d-form store): Likewise.
22503 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22506 * config/i386/mmx.md (*mov<mode>_internal): Enable
22507 alternatives 11, 12, 13 and 14 also for 32bit targets.
22508 Remove alternatives 15, 16, 17 and 18.
22509 * config/i386/sse.md (vec_concatv2di): Change
22510 alternative (!x, *y) to (x, ?!*Yn).
22512 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22514 * dumpfile.h (enum dump_kind): Remove stray comma.
22516 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22518 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22519 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22520 predicate::num_conditions
22521 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22522 (CHANGED): turn into predicate::changed.
22523 (agg_position_info): Move to ipa-predicate.h
22524 (add_condition, predicate::add_clause, predicate::operator &=,
22525 predicate::or_with, predicate::evaluate, predicate::probability,
22526 dump_condition, dump_clause, predicate::dump,
22527 predicate::remap_after_duplication, predicate::remap_after_inlining,
22528 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22529 (evaluate_conditions_for_known_args): Update.
22530 (set_cond_stmt_execution_predicate): Update.
22531 * ipa-inline.h: Include ipa-predicate.h
22532 (condition, inline_param_summary, conditions, agg_position_info,
22533 predicate): Move to ipa-predicate.h
22534 * ipa-predicate.c: New file.
22535 * ipa-predicate.h: New file.
22537 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22539 * final.c (leaf_function_p): Check we are not in a sequence.
22541 2017-05-18 Martin Liska <mliska@suse.cz>
22543 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22544 * dumpfile.c (dump_register): Use new enum dump_kind.
22545 (get_dump_file_name): Likewise.
22546 (dump_enable_all): Likewise.
22547 (dump_switch_p_1): Likewise.
22548 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22549 * dumpfile.h (enum dump_kind): New enum type.
22550 (struct dump_file_info): Create constructor and
22551 format fields and comments.
22552 * passes.c (pass_manager::register_one_dump_file):
22554 * statistics.c (statistics_early_init): Likewise.
22555 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22556 TDF_TREE with TDF_SLIM.
22557 (gather_memory_references_ref): Likewise.
22559 2017-05-18 Martin Liska <mliska@suse.cz>
22561 * vec.h (struct vnull): Use it.
22563 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22565 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22566 (true_predicate, false_predicate, true_predicate_p,
22567 false_predicate_p): Remove.
22568 (single_cond_predicate, not_inlined_predicate): Turn to member function
22570 (add_condition): Update.
22571 (add_clause): Turn to...
22572 (predicate::add_clause): ... this one; update; allow passing NULL
22574 (and_predicates): Turn to ...
22575 (predicate::operator &=): ... this one.
22576 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22577 (or_predicates): Turn to ...
22578 (predicate::or_with): ... this one.
22579 (evaluate_predicate): Turn to ...
22580 (predicate::evaluate): ... this one.
22581 (predicate_probability): Turn to ...
22582 (predicate::probability): ... this one.
22583 (dump_condition): Update.
22584 (dump_predicate): Turn to ...
22585 (predicate::dump): ... this one.
22586 (account_size_time): Update.
22587 (edge_set_predicate): Update.
22588 (set_hint_predicate): UPdate.
22589 (evaluate_conditions_for_known_args): Update.
22590 (evaluate_properties_for_edge): Update.
22591 (remap_predicate_after_duplication): Turn to...
22592 (predicate::remap_after_duplication): ... this one.
22593 (remap_hint_predicate_after_duplication): Update.
22594 (inline_summary_t::duplicate): UPdate.
22595 (dump_inline_edge_summary): Update.
22596 (dump_inline_summary): Update.
22597 (set_cond_stmt_execution_predicate): Update.
22598 (set_switch_stmt_execution_predicate): Update.
22599 (compute_bb_predicates): Update.
22600 (will_be_nonconstant_expr_predicate): Update.
22601 (will_be_nonconstant_predicate): Update.
22602 (phi_result_unknown_predicate): Update.
22603 (predicate_for_phi_result): Update.
22604 (array_index_predicate): Update.
22605 (estimate_function_body_sizes): Update.
22606 (estimate_node_size_and_time): Update.
22607 (estimate_ipcp_clone_size_and_time): Update.
22608 (remap_predicate): Rename to ...
22609 (predicate::remap_after_inlining): ... this one.
22610 (remap_hint_predicate): Update.
22611 (inline_merge_summary): Update.
22612 (inline_update_overall_summary): Update.
22613 (estimate_size_after_inlining): Update.
22614 (read_predicate): Rename to ...
22615 (predicate::stream_in): ... this one.
22616 (read_inline_edge_summary): Update.
22617 (write_predicate): Rename to ...
22618 (predicate::stream_out): ... this one.
22619 (write_inline_edge_summary): Update.
22620 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22621 (clause_t): Turn to uint32_t
22622 (predicate): Turn to class; implement constructor and operators
22624 (size_time_entry): Update.
22625 (inline_summary): Update.
22626 (inline_edge_summary): Update.
22628 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22630 * fold-const.c (fold_binary_loc): Move transformation...
22631 * match.pd (C - X CMP X): ... here.
22633 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22635 * config/sparc/sparc.c (sparc_option_override): Set function
22636 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22637 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22639 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22641 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22643 2017-05-18 Marek Polacek <polacek@redhat.com>
22646 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22647 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22648 gimple_assign_load_p.
22650 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22652 PR middle-end/80692
22653 * real.c (do_compare): Give decimal_do_compare preference over
22654 comparing just the signs.
22656 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22658 * doc/md.texi (Canonicalization of Instructions): Describe the
22659 canonical form of instructions that inherently set a condition
22662 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22664 PR middle-end/80775
22665 * tree-cfg.c: Move deletion of unreachable case statements to after
22666 the merging of consecutive case labels.
22668 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22670 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22671 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22672 restoring of callee-saved registers.
22674 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22676 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22677 * config/visium/visium.c (single_set_and_flags): Likewise.
22678 * config/visium/visium.md (Substitutions): Likewise.
22680 2017-05-17 Martin Liska <mliska@suse.cz>
22682 * cfg.c: Introduce dump_flags_t type and
22683 use it instead of int type.
22685 * cfghooks.c: Likewise.
22686 * cfghooks.h (struct cfg_hooks): Likewise.
22687 * cfgrtl.c: Likewise.
22688 * cfgrtl.h: Likewise.
22689 * cgraph.c (cgraph_node::get_body): Likewise.
22690 * coretypes.h: Likewise.
22691 * domwalk.c: Likewise.
22692 * domwalk.h: Likewise.
22693 * dumpfile.c (struct dump_option_value_info): Likewise.
22694 (dump_enable_all): Likewise.
22695 (dump_switch_p_1): Likewise.
22696 (opt_info_switch_p): Likewise.
22697 * dumpfile.h (enum tree_dump_index): Likewise.
22698 (struct dump_file_info): Likewise.
22699 * genemit.c: Likewise.
22700 * generic-match-head.c: Likewise.
22701 * gengtype.c (open_base_files): Likewise.
22702 * gimple-pretty-print.c: Likewise.
22703 * gimple-pretty-print.h: Likewise.
22704 * graph.c (print_graph_cfg): Likewise.
22705 * graphite-scop-detection.c (dot_all_sese): Likewise.
22706 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22707 * loop-unroll.c (report_unroll): Likewise.
22708 * passes.c (pass_manager::register_one_dump_file): Likewise.
22709 * print-tree.c: Likewise.
22710 * statistics.c: Likewise.
22711 * tree-cfg.c: Likewise.
22712 * tree-cfg.h: Likewise.
22713 * tree-dfa.c: Likewise.
22714 * tree-dfa.h: Likewise.
22715 * tree-dump.c (dump_function): Likewise.
22716 * tree-dump.h (struct dump_info): Likewise.
22717 * tree-pretty-print.c: Likewise.
22718 * tree-pretty-print.h: Likewise.
22719 * tree-ssa-live.c: Likewise.
22720 * tree-ssa-live.h: Likewise.
22721 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22722 * tree-vect-loop.c: Likewise.
22723 * tree-vect-slp.c: Likewise.
22725 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22726 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22728 PR tree-optimization/80457
22729 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22730 of all arguments to a statement as scalar_to_vec operations.
22731 (vectorizable_call): Adjust call to vect_model_simple_cost for
22733 (vectorizable_conversion): Likewise.
22734 (vectorizable_assignment): Likewise.
22735 (vectorizable_shift): Likewise.
22736 (vectorizable_operation): Likewise.
22737 (vectorizable_comparison): Likewise.
22738 (vect_is_simple_cond): Record the def types for operands.
22739 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22740 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22741 for statement argument count.
22743 2017-05-16 Carl Love <cel@us.ibm.com>
22745 * config/rs6000/rs6000-c: Add support for built-in functions
22746 vector unsigned long long vec_bperm (vector unsigned long long,
22747 vector unsigned char)
22748 vector signed long long vec_mule (vector signed int,
22750 vector unsigned long long vec_mule (vector unsigned int,
22751 vector unsigned int)
22752 vector signed long long vec_mulo (vector signed int,
22754 vector unsigned long long vec_mulo (vector unsigned int,
22755 vector unsigned int)
22756 vector signed char vec_sldw (vector signed char,
22757 vector signed char,
22759 vector unsigned char vec_sldw (vector unsigned char,
22760 vector unsigned char,
22762 vector signed short vec_sldw (vector signed short,
22763 vector signed short,
22765 vector unsigned short vec_sldw (vector unsigned short,
22766 vector unsigned short,
22768 vector signed int vec_sldw (vector signed int,
22771 vector unsigned int vec_sldw (vector unsigned int,
22772 vector unsigned int,
22774 vector signed long long vec_sldw (vector signed long long,
22775 vector signed long long,
22777 vector unsigned long long vec_sldw (vector unsigned long long,
22778 vector unsigned long long,
22780 * config/rs6000/rs6000-c: Add support for built-in functions
22781 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22782 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22783 * doc/extend.texi: Update the built-in documentation for the
22784 new built-in functions.
22786 2017-05-16 Marek Polacek <polacek@redhat.com>
22790 * tree.c (save_expr): Don't fold the expression.
22792 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22794 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22795 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22796 and (?*y,m). Update insn attributes.
22798 2017-05-16 Martin Liska <mliska@suse.cz>
22800 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22801 flags argument of print_gimple_stmt, print_gimple_expr,
22802 print_generic_stmt and print_generic_expr.
22803 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22804 * coretypes.h: Likewise.
22805 * except.c (dump_eh_tree): Likewise.
22806 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22807 * gimple-pretty-print.h: Likewise.
22808 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22809 (backprop::push_to_worklist): Likewise.
22810 (backprop::pop_from_worklist): Likewise.
22811 (backprop::process_use): Likewise.
22812 (backprop::intersect_uses): Likewise.
22813 (note_replacement): Likewise.
22814 * gimple-ssa-store-merging.c
22815 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22816 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22817 (pass_store_merging::execute): Likewise.
22818 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22819 (ssa_base_cand_dump_callback): Likewise.
22820 (dump_incr_vec): Likewise.
22821 (replace_refs): Likewise.
22822 (replace_mult_candidate): Likewise.
22823 (create_add_on_incoming_edge): Likewise.
22824 (create_phi_basis): Likewise.
22825 (insert_initializers): Likewise.
22826 (all_phi_incrs_profitable): Likewise.
22827 (introduce_cast_before_cand): Likewise.
22828 (replace_one_candidate): Likewise.
22829 * gimplify.c (gimplify_expr): Likewise.
22830 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22831 (set_rename): Likewise.
22832 (rename_uses): Likewise.
22833 (copy_loop_phi_nodes): Likewise.
22834 (add_close_phis_to_merge_points): Likewise.
22835 (copy_loop_close_phi_args): Likewise.
22836 (copy_cond_phi_args): Likewise.
22837 (graphite_copy_stmts_from_block): Likewise.
22838 (translate_pending_phi_nodes): Likewise.
22839 * graphite-poly.c (print_pdr): Likewise.
22840 (dump_gbb_cases): Likewise.
22841 (dump_gbb_conditions): Likewise.
22842 (print_scop_params): Likewise.
22843 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22844 (build_cross_bb_scalars_use): Likewise.
22845 (gather_bbs::before_dom_children): Likewise.
22846 * hsa-dump.c (dump_hsa_immed): Likewise.
22847 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22848 (get_replacement_map): Likewise.
22849 * ipa-inline-analysis.c (dump_condition): Likewise.
22850 (estimate_function_body_sizes): Likewise.
22851 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22852 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22853 * ipa-prop.c (ipa_dump_param): Likewise.
22854 (ipa_print_node_jump_functions_for_edge): Likewise.
22855 (ipa_modify_call_arguments): Likewise.
22856 (ipa_modify_expr): Likewise.
22857 (ipa_dump_param_adjustments): Likewise.
22858 (ipa_dump_agg_replacement_values): Likewise.
22859 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22860 * ipa-pure-const.c (check_stmt): Likewise.
22861 (pass_nothrow::execute): Likewise.
22862 * ipa-split.c (execute_split_functions): Likewise.
22863 * omp-offload.c (dump_oacc_loop_part): Likewise.
22864 (dump_oacc_loop): Likewise.
22865 * trans-mem.c (tm_log_emit): Likewise.
22866 (tm_memopt_accumulate_memops): Likewise.
22867 (dump_tm_memopt_set): Likewise.
22868 (dump_tm_memopt_transform): Likewise.
22869 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22870 (print_loop): Likewise.
22871 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22872 (chkp_gather_checks_info): Likewise.
22873 (chkp_get_check_result): Likewise.
22874 (chkp_remove_check_if_pass): Likewise.
22875 (chkp_use_outer_bounds_if_possible): Likewise.
22876 (chkp_reduce_bounds_lifetime): Likewise.
22877 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22878 (chkp_mark_completed_bounds): Likewise.
22879 (chkp_register_incomplete_bounds): Likewise.
22880 (chkp_mark_invalid_bounds): Likewise.
22881 (chkp_maybe_copy_and_register_bounds): Likewise.
22882 (chkp_build_returned_bound): Likewise.
22883 (chkp_get_bound_for_parm): Likewise.
22884 (chkp_build_bndldx): Likewise.
22885 (chkp_get_bounds_by_definition): Likewise.
22886 (chkp_generate_extern_var_bounds): Likewise.
22887 (chkp_get_bounds_for_decl_addr): Likewise.
22888 * tree-chrec.c (chrec_apply): Likewise.
22889 * tree-data-ref.c (dump_data_reference): Likewise.
22890 (dump_subscript): Likewise.
22891 (dump_data_dependence_relation): Likewise.
22892 (analyze_overlapping_iterations): Likewise.
22893 * tree-inline.c (expand_call_inline): Likewise.
22894 (tree_function_versioning): Likewise.
22895 * tree-into-ssa.c (dump_defs_stack): Likewise.
22896 (dump_currdefs): Likewise.
22897 (dump_names_replaced_by): Likewise.
22898 (dump_update_ssa): Likewise.
22899 (update_ssa): Likewise.
22900 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22901 * tree-parloops.c (build_new_reduction): Likewise.
22902 (try_create_reduction_list): Likewise.
22903 (ref_conflicts_with_region): Likewise.
22904 (oacc_entry_exit_ok_1): Likewise.
22905 (oacc_entry_exit_single_gang): Likewise.
22906 * tree-pretty-print.h: Likewise.
22907 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22908 (get_scalar_evolution): Likewise.
22909 (add_to_evolution): Likewise.
22910 (get_loop_exit_condition): Likewise.
22911 (analyze_evolution_in_loop): Likewise.
22912 (analyze_initial_condition): Likewise.
22913 (analyze_scalar_evolution): Likewise.
22914 (instantiate_scev): Likewise.
22915 (number_of_latch_executions): Likewise.
22916 (gather_chrec_stats): Likewise.
22917 (final_value_replacement_loop): Likewise.
22918 (scev_const_prop): Likewise.
22919 * tree-sra.c (dump_access): Likewise.
22920 (disqualify_candidate): Likewise.
22921 (create_access): Likewise.
22922 (reject): Likewise.
22923 (maybe_add_sra_candidate): Likewise.
22924 (create_access_replacement): Likewise.
22925 (analyze_access_subtree): Likewise.
22926 (analyze_all_variable_accesses): Likewise.
22927 (sra_modify_assign): Likewise.
22928 (initialize_constant_pool_replacements): Likewise.
22929 (find_param_candidates): Likewise.
22930 (decide_one_param_reduction): Likewise.
22931 (replace_removed_params_ssa_names): Likewise.
22932 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22933 * tree-ssa-copy.c (dump_copy_of): Likewise.
22934 (copy_prop_visit_cond_stmt): Likewise.
22935 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22936 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22937 (record_equivalences_from_stmt): Likewise.
22938 * tree-ssa-dse.c (compute_trims): Likewise.
22939 (delete_dead_call): Likewise.
22940 (delete_dead_assignment): Likewise.
22941 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22942 (forward_propagate_into_cond): Likewise.
22943 (pass_forwprop::execute): Likewise.
22944 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22945 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22947 (move_computations_worker): Likewise.
22948 (execute_sm): Likewise.
22949 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22950 (remove_exits_and_undefined_stmts): Likewise.
22951 (remove_redundant_iv_tests): Likewise.
22952 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22953 (adjust_iv_update_pos): Likewise.
22954 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22955 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22956 (value_replacement): Likewise.
22957 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22958 * tree-ssa-pre.c (print_pre_expr): Likewise.
22959 (get_representative_for): Likewise.
22960 (create_expression_by_pieces): Likewise.
22961 (insert_into_preds_of_block): Likewise.
22962 (eliminate_insert): Likewise.
22963 (eliminate_dom_walker::before_dom_children): Likewise.
22964 (eliminate): Likewise.
22965 (remove_dead_inserted_code): Likewise.
22966 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22967 * tree-ssa-reassoc.c (get_rank): Likewise.
22968 (eliminate_duplicate_pair): Likewise.
22969 (eliminate_plus_minus_pair): Likewise.
22970 (eliminate_not_pairs): Likewise.
22971 (undistribute_ops_list): Likewise.
22972 (eliminate_redundant_comparison): Likewise.
22973 (update_range_test): Likewise.
22974 (optimize_range_tests_var_bound): Likewise.
22975 (optimize_vec_cond_expr): Likewise.
22976 (rewrite_expr_tree): Likewise.
22977 (rewrite_expr_tree_parallel): Likewise.
22978 (linearize_expr): Likewise.
22979 (break_up_subtract): Likewise.
22980 (linearize_expr_tree): Likewise.
22981 (attempt_builtin_powi): Likewise.
22982 (attempt_builtin_copysign): Likewise.
22983 (transform_stmt_to_copy): Likewise.
22984 (transform_stmt_to_multiply): Likewise.
22985 (dump_ops_vector): Likewise.
22986 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22987 (print_scc): Likewise.
22988 (set_ssa_val_to): Likewise.
22989 (visit_reference_op_store): Likewise.
22990 (visit_use): Likewise.
22991 (sccvn_dom_walker::before_dom_children): Likewise.
22992 (run_scc_vn): Likewise.
22993 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22995 (expr_hash_elt::print): Likewise.
22996 (const_and_copies::pop_to_marker): Likewise.
22997 (const_and_copies::record_const_or_copy_raw): Likewise.
22998 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22999 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23000 (dump_predicates): Likewise.
23001 (find_uninit_use): Likewise.
23002 (warn_uninitialized_phi): Likewise.
23003 (pass_late_warn_uninitialized::execute): Likewise.
23004 * tree-ssa.c (verify_vssa): Likewise.
23005 (verify_ssa): Likewise.
23006 (maybe_optimize_var): Likewise.
23007 * tree-vrp.c (dump_value_range): Likewise.
23008 (dump_all_value_ranges): Likewise.
23009 (dump_asserts_for): Likewise.
23010 (register_edge_assert_for_2): Likewise.
23011 (vrp_visit_cond_stmt): Likewise.
23012 (vrp_visit_switch_stmt): Likewise.
23013 (vrp_visit_stmt): Likewise.
23014 (vrp_visit_phi_node): Likewise.
23015 (simplify_cond_using_ranges_1): Likewise.
23016 (fold_predicate_in): Likewise.
23017 (evrp_dom_walker::before_dom_children): Likewise.
23018 (evrp_dom_walker::push_value_range): Likewise.
23019 (evrp_dom_walker::pop_value_range): Likewise.
23020 (execute_early_vrp): Likewise.
23022 2017-05-16 Richard Biener <rguenther@suse.de>
23024 * dwarf2out.c (loc_list_from_tree_1): Do not create
23025 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23027 2017-05-16 Richard Biener <rguenther@suse.de>
23029 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23031 (note_variable_value_in_expr): If we resolved the decl ref
23032 do not push to the stack.
23034 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23036 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23037 operations in fast-math mode.
23038 (vaddq_f16): Likewise.
23039 (vmul_f16): Likewise.
23040 (vmulq_f16): Likewise.
23041 (vsub_f16): Likewise.
23042 (vsubq_f16): Likewise.
23043 * config/arm/neon.md (add<mode>3): New.
23045 (fma:<VH:mode>3): New. Also remove outdated comment.
23048 2017-05-16 Martin Liska <mliska@suse.cz>
23052 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23053 (odr_types_equivalent_p): Likewise.
23055 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23057 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23059 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23062 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23063 non-interunit SSE move alternatives with '?'.
23064 (zero-extendsidi peephole2): New peephole to skip intermediate
23065 general register in SSE zero-extend sequence.
23067 2017-05-15 Jeff Law <law@redhat.com>
23069 * reorg.c (relax_delay_slots): Create a new variable to hold
23070 the temporary target rather than clobbering TARGET_LABEL.
23072 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23073 missing argument to extract_bit_field call.
23074 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23076 2017-05-15 Martin Liska <mliska@suse.cz>
23079 * gcc.c (process_command): Do not allow empty argument of -o option.
23081 2017-05-15 Renlin Li <renlin.li@arm.com>
23083 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23084 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23085 * config/aarch64/constraints.md (Usf): Add long call check.
23086 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23087 (call_value): Likewise.
23088 (sibcall): Likewise.
23089 (sibcall_value): Likewise.
23091 (call_value_insn): New.
23092 (sibcall_insn): Update rtx pattern.
23093 (sibcall_value_insn): Likewise.
23094 (call_internal): Remove.
23095 (call_value_internal): Likewise.
23096 (sibcall_internal): Likewise.
23097 (sibcall_value_internal): Likewise.
23098 (call_reg): Likewise.
23099 (call_symbol): Likewise.
23100 (call_value_reg): Likewise.
23101 (call_value_symbol): Likewise.
23103 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23106 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23108 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23110 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23111 compatible with CCGOCmode and with CCZmode.
23113 2017-05-14 Martin Sebor <msebor@redhat.com>
23115 PR middle-end/77671
23116 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23117 (gimple_fold_builtin_snprintf): Same.
23118 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23119 (gimple_fold_builtin_snprintf): Same.
23120 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23121 of character types.
23122 (is_call_safe): New function.
23123 (try_substitute_return_value): Call it.
23124 (try_simplify_call): New function.
23125 (pass_sprintf_length::handle_gimple_call): Call it.
23127 2017-05-14 Martin Sebor <msebor@redhat.com>
23129 PR middle-end/80669
23130 * builtins.c (expand_builtin_stpncpy): Simplify.
23132 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23134 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23135 * config/i386/i386.h
23136 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23137 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23138 (struct machine_function): Add new members call_ms2sysv,
23139 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23140 (struct machine_frame_state): New fields sp_realigned and
23141 sp_realigned_offset.
23142 * config/i386/i386.c
23143 (enum xlogue_stub): New enum.
23144 (enum xlogue_stub_sets): New enum.
23145 (class xlogue_layout): New class.
23146 (struct ix86_frame): New fields stack_realign_allocate_offset,
23147 stack_realign_offset and outlined_save_offset. Modify comments to
23148 detail stack layout when using out-of-line stubs.
23149 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23150 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23151 -mcall-ms2sysv-xlogues.
23152 (stub_managed_regs): New static variable.
23153 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23154 registers managed by out-of-line stub.
23155 (disable_call_ms2sysv_xlogues): New function.
23156 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23157 m->call_ms2sysv when appropriate and compute frame layout for
23159 (sp_valid_at, fp_valid_at): New inline functions.
23160 (choose_basereg): New function.
23161 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23163 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23164 Use align parameter of choose_baseaddr to generated aligned SSE movs
23166 (pro_epilogue_adjust_stack): Modify to track
23167 machine_frame_state::sp_realigned.
23168 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23169 (ix86_nsaved_sseregs): Likewise.
23170 (ix86_emit_save_regs): Likewise.
23171 (ix86_emit_save_regs_using_mov): Likewise.
23172 (ix86_emit_save_sse_regs_using_mov): Likewise.
23173 (get_scratch_register_on_entry): Likewise.
23174 (gen_frame_set): New function.
23175 (gen_frame_load): Likewise.
23176 (gen_frame_store): Likewise.
23177 (emit_outlined_ms2sysv_save): Likewise.
23178 (emit_outlined_ms2sysv_restore): Likewise.
23179 (ix86_expand_prologue): Modify stack re-alignment code and call
23180 emit_outlined_ms2sysv_save when appropriate.
23181 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23182 parameter rtx_insn *insn, which allows the function to be used to only
23183 generate the notes.
23184 (ix86_expand_epilogue): Modify validity checks of frame and stack
23185 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23186 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23187 * config/i386/predicates.md
23188 (save_multiple): New predicate.
23189 (restore_multiple): Likewise.
23190 * config/i386/sse.md
23191 (save_multiple<mode>): New pattern.
23192 (save_multiple_realign<mode>): Likewise.
23193 (restore_multiple<mode>): Likewise.
23194 (restore_multiple_and_return<mode>): Likewise.
23195 (restore_multiple_leave_return<mode>): Likewise.
23196 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23198 2017-05-14 Julia Koval <julia.koval@intel.com>
23200 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23201 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23202 (__builtin_ia32_xsetbv): New builtins.
23203 * config/i386/i386.c (ix86_expand_special_args_builtin):
23205 (ix86_expand_builtin): Special expand for new intrinsics.
23206 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23207 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23208 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23210 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23212 * cfganal.c (inverted_post_order_compute): Change argument type
23214 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23215 * df-core.c (rest_of_handle_df_initialize): Adjust.
23216 (rest_of_handle_df_finish): Likewise.
23217 (df_analyze_1): Likewise.
23218 (df_analyze): Likewise.
23219 (loop_inverted_post_order_compute): Change argument to be a vec *.
23220 (df_analyze_loop): Adjust.
23221 (df_get_n_blocks): Likewise.
23222 (df_get_postorder): Likewise.
23223 * df.h (struct df_d): Change field to be a vec.
23224 * lcm.c (compute_laterin): Adjust.
23225 (compute_available): Likewise.
23226 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23227 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23228 * tree-ssa-pre.c (compute_antic): Likewise.
23230 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23232 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23233 (depth_first_search::depth_first_search): Change structure init
23234 function to this constructor.
23235 (depth_first_search::add_bb): Rename function to this member.
23236 (depth_first_search::execute): Likewise.
23237 (flow_dfs_compute_reverse_finish): Adjust.
23239 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23241 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23242 (longest_simple_path): Likewise.
23243 * shrink-wrap.c (spread_components): Likewise.
23244 (disqualify_problematic_components): Likewise.
23245 (emit_common_heads_for_components): Likewise.
23246 (emit_common_tails_for_components): Likewise.
23247 (insert_prologue_epilogue_for_components): Likewise.
23249 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23251 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23254 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23256 * df-core.c (df_set_blocks): Start using auto_bitmap.
23257 (df_compact_blocks): Likewise.
23258 * df-problems.c (df_rd_confluence_n): Likewise.
23259 * df-scan.c (df_insn_rescan_all): Likewise.
23260 (df_process_deferred_rescans): Likewise.
23261 (df_update_entry_block_defs): Likewise.
23262 (df_update_exit_block_uses): Likewise.
23263 (df_entry_block_bitmap_verify): Likewise.
23264 (df_exit_block_bitmap_verify): Likewise.
23265 (df_scan_verify): Likewise.
23266 * lra-constraints.c (lra_constraints): Likewise.
23267 (undo_optional_reloads): Likewise.
23268 (lra_undo_inheritance): Likewise.
23269 * lra-remat.c (calculate_gen_cands): Likewise.
23270 (do_remat): Likewise.
23271 * lra-spills.c (assign_spill_hard_regs): Likewise.
23272 (spill_pseudos): Likewise.
23273 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23274 (bitmap_set_subtract_values): Likewise.
23276 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23278 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23279 management with auto_bitmap.
23280 (fix_inter_tick): Likewise.
23281 (fix_recovery_deps): Likewise.
23282 * ira.c (add_store_equivs): Likewise.
23283 (find_moveable_pseudos): Likewise.
23284 (split_live_ranges_for_shrink_wrap): Likewise.
23285 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23286 (rtx_reuse_manager::seen_def_p): Likewise.
23287 (rtx_reuse_manager::set_seen_def): Likewise.
23288 * print-rtl.h (class rtx_reuse_manager): Likewise.
23290 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23292 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23294 (migrate_btr_def): Likewise.
23295 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23296 * df-core.c (loop_post_order_compute): Likewise.
23297 (loop_inverted_post_order_compute): Likewise.
23298 * hsa-common.h: Likewise.
23299 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23300 * init-regs.c (initialize_uninitialized_regs): Likewise.
23301 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23302 (inline_small_functions): Likewise.
23303 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23304 * ira.c (combine_and_move_insns): Likewise.
23305 (build_insn_chain): Likewise.
23306 * loop-invariant.c (find_invariants): Likewise.
23307 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23308 * predict.c (tree_predict_by_opcode): Likewise.
23309 (predict_paths_leading_to): Likewise.
23310 (predict_paths_leading_to_edge): Likewise.
23311 (estimate_loops_at_level): Likewise.
23312 (estimate_loops): Likewise.
23313 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23314 (spread_components): Likewise.
23315 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23316 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23317 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23318 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23319 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23320 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23321 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23322 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23323 (mark_threaded_blocks): Likewise.
23324 (thread_through_all_blocks): Likewise.
23325 * tree-ssa.c (verify_ssa): Likewise.
23326 (execute_update_addresses_taken): Likewise.
23327 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23329 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23331 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23333 (post_order_compute): Likewise.
23334 (inverted_post_order_compute): Likewise.
23335 (pre_and_rev_post_order_compute_fn): Likewise.
23337 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23339 * genrecog.c (int_set::int_set): Explicitly construct our
23340 auto_vec base class.
23341 * vec.h (auto_vec::auto_vec): New constructor.
23343 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23345 * bitmap.h (class auto_bitmap): New constructor taking
23346 bitmap_obstack * argument.
23348 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23350 * bitmap.h (class auto_bitmap): Change type of m_bits to
23351 bitmap_head, and adjust ctor / dtor and member operators.
23353 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23355 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23356 when returned register mode doesn't match original mode.
23358 2017-05-12 Jeff Law <law@redhat.com>
23359 Jakub Jelinek <jakub@redhat.com>
23361 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23362 we look for cc setter after the compare-elim changes.
23363 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23364 within the vector to match what compare-elim now expects.
23365 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23366 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23368 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23369 after the compare-elim changes.
23370 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23371 the vector to match what compare-elim now expects.
23372 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23373 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23374 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23375 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23376 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23378 * config/visium/visium.c (single_set_and_flags): Fix where
23379 we look for cc setter after the compare-elim changes.
23380 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23381 with the vector to match what compare-elim now expects.
23382 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23383 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23384 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23385 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23386 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23387 (neg<mode>2_insn_set_overflow): Likewise.
23389 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23391 PR middle-end/79794
23392 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23393 maybe_expand_insn call, set ops[0].target. If still set after call,
23394 set alt_rtl. Add extra arg to recursive calls.
23395 (extract_bit_field): Add alt_rtl argument. Pass to
23397 * expmed.h (extract_bit_field): Fix prototype.
23398 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23399 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23400 to extract_bit_field_calls.
23401 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23402 Pass alt_rtl to extract_bit_field calls.
23403 * calls.c (store_unaligned_arguments_into_psuedos)
23404 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23405 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23407 * optabs.h (struct expand_operand): Add target bitfield.
23409 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23411 * compare-elim.c (try_eliminate_compare): Canonicalize
23412 operation with embedded compare to
23413 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23414 (set (reg) (operation)].
23416 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23418 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23421 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23422 cost of adding a carry flag for ADC instruction.
23423 [case MINUS]: Ignore the cost of subtracting a carry flag
23424 for SBB instruction.
23426 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23428 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23430 * config/rs6000/bmiintrin.h: New file.
23431 * config/rs6000/bmi2intrin.h: New file.
23432 * config/rs6000/x86intrin.h: New file.
23434 2017-05-12 Jeff Law <law@redhat.com>
23436 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23439 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23441 PR middle-end/80707
23442 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23444 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23446 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23447 early expansion of vector divide builtins.
23448 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23449 builtins identified as having unsigned arguments.
23451 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23453 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23454 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23455 expansion of vector logical operations (and, andc, or, xor,
23458 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23460 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23461 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23463 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23465 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23466 early GIMPLE expansion of vector multiplies.
23468 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23470 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23471 TARGET_HAVE_MOVT conditional.
23472 (movt splitter): Likewise.
23474 2017-05-12 Richard Biener <rguenther@suse.de>
23476 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23477 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23478 Fold all stmts not inplace.
23480 2017-05-12 Richard Biener <rguenther@suse.de>
23482 PR tree-optimization/80713
23483 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23484 inserted_exprs bit for not removed stmts.
23486 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23488 PR middle-end/69921
23489 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23490 parallelized" attribute for parallelized OpenACC kernels.
23491 * omp-offload.c (execute_oacc_device_lower): Use it.
23493 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23494 Set "oacc kernels" attribute.
23495 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23496 parameter. Adjust all users.
23497 (oacc_fn_attrib_kernels_p): Remove function.
23498 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23499 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23500 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23501 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23502 assert "oacc kernels" attribute is set.
23504 2017-05-11 Carl Love <cel@us.ibm.com>
23506 * config/rs6000/rs6000-c: Add support for built-in functions
23507 vector unsigned char vec_popcnt (vector signed char)
23508 vector unsigned char vec_popcnt (vector unsigned char)
23509 vector unsigned short vec_popcnt (vector signed short)
23510 vector unsigned short vec_popcnt (vector unsigned short)
23511 vector unsigned int vec_popcnt (vector signed int)
23512 vector unsigned int vec_popcnt (vector unsigned int)
23513 vector unsigned long long vec_popcnt (vector signed long long)
23514 vector unsigned long long vec_popcnt (vector unsigned long long)
23515 vector signed long long vec_slo (vector signed long long,
23516 vector signed char)
23517 vector signed long long vec_slo (vector signed long long,
23518 vector unsigned char)
23519 vector unsigned long long vec_slo (vector unsigned long long,
23520 vector signed char)
23521 vector unsigned long long vec_slo (vector unsigned long long,
23522 vector unsigned char)
23523 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23524 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23525 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23526 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23527 * doc/extend.texi: Update the built-in documentation file for the
23528 new built-in functions.
23530 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23532 * attribs.h (sorted_attr_string): Move machine independent
23533 functions for target clone support from the i386 port to common
23534 code. Rename ix86_function_versions to common_function_versions.
23535 Rename make_name to make_unique_name.
23536 (common_function_versions): Likewise.
23537 (make_unique_name): Likewise.
23538 (make_dispatcher_decl): Likewise.
23539 (is_function_default_version): Likewise.
23540 * attribs.c (attr_strcmp): Likewise.
23541 (sorted_attr_string): Likewise.
23542 (common_function_versions): Likewise.
23543 (make_unique_name): Likewise.
23544 (make_dispatcher_decl): Likewise.
23545 (is_function_default_version): Likewise.
23546 * config/i386/i386.c (attr_strcmp): Likewise.
23547 (sorted_attr_string): Likewise.
23548 (ix86_function_versions): Likewise.
23549 (make_name): Likewise.
23550 (make_dispatcher_decl): Likewise.
23551 (is_function_default_version): Likewise.
23552 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23554 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23557 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23558 Account for direct move costs for vec_construct of integer
23561 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23564 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23565 (UNSPEC_STX_ATOMIC): Ditto.
23566 (loaddi_via_sse): New insn.
23567 (storedi_via_sse): Ditto.
23568 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23569 Update corresponding peephole2 patterns.
23570 (atomic_storedi_fpu): Ditto.
23572 2017-05-11 Julia Koval <julia.koval@intel.com>
23574 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23575 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23577 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23578 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23579 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23581 2017-05-11 Nathan Sidwell <nathan@acm.org>
23583 * graphite-poly.c: Include dumpfile.h.
23585 * dumpfle.h (dump_function): Declare here ...
23586 * tree-dump.h (dump_function): ... not here.
23587 * dumpfile.c: #include tree-cfg.h.
23588 (dump_function): Move here from ...
23589 * tree-dump.c (dump_function): ... here.
23590 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23591 * graphite-poly.c: Don't include tree-dump.h.
23592 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23593 * print-tree.c: Likewise.
23594 * stor-layout.c: Likewise.
23595 * tree-nested.c: Likewise.
23597 * dumpfile.c (dump_start): Use TDF_FLAGS.
23598 (dump_enable_all): Fix TDF_KIND check thinko.
23600 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23602 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23603 array entries to represent two legal parameterizations of the
23604 overloaded __builtin_cmpb function, as represented by the
23605 P6_OV_BUILTIN_CMPB constant.
23606 (altivec_resolve_overloaded_builtin): Add special case handling
23607 for the __builtin_cmpb function, as represented by the
23608 P6_OV_BUILTIN_CMPB constant.
23609 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23610 (BU_P6_64BIT_2): New macro.
23611 (BU_P6_OVERLOAD_2): New macro
23612 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23613 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23614 (CMPB): Add overload support to represent both 32-bit and 64-bit
23615 compare-bytes function.
23616 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23617 support for TARGET_CMPB.
23618 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23619 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23620 documentation of the __builtin_cmpb overloaded built-in function.
23622 2017-05-11 Richard Biener <rguenther@suse.de>
23624 PR tree-optimization/80705
23625 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23626 bases are not vectorizable.
23628 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23630 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23631 when counting register pressure.
23633 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23635 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23636 (iv_ca_more_deps): Renamed to ...
23637 (iv_ca_compare_deps): ... this.
23638 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23642 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23644 (determine_group_iv_costs): ... here.
23645 (find_inv_vars_cb): Record inv var if it's not recorded before.
23647 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23649 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23650 (get_shiftadd_cost): Ditto.
23652 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23654 * tree-ssa-address.c: Include header file.
23655 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23657 (add_to_parts): Refactor.
23658 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23659 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23662 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23664 PR tree-optimization/53090
23665 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23667 (extract_cond_operands): Detect condition with IV on both sides
23668 and return COMP_IV_EXPR_2.
23669 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23670 (rewrite_use_compare): Simplify by removing call to function
23671 extract_cond_operands.
23673 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23675 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23676 (extract_cond_operands): Detect condition comparing against non-
23677 invariant bound and return appropriate enum value.
23678 (find_interesting_uses_cond): Update use of extract_cond_operands.
23679 Handle its return value accordingly.
23680 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23682 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23684 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23685 nonlinear iv_use computation in loop invariant sensitive way.
23687 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23689 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23690 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23692 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23694 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23695 (dump_cand): Support iv_cand.inv_exprs.
23696 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23698 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23701 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23703 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23705 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23706 as local function. Include necessary header files.
23707 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23709 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23711 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23713 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23715 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23716 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23717 RSHIFT_EXPR and BIT_NOT_EXPR.
23719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23721 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23722 (adjust_setup_cost): New parameter supporting round up adjustment.
23723 (struct address_cost_data): Delete.
23724 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23725 (split_address_cost, ptr_difference_cost): Delete.
23726 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23727 (struct ainc_cost_data): New struct.
23728 (get_address_cost_ainc): New function.
23729 (get_address_cost, get_computation_cost): Reimplement.
23730 (determine_group_iv_cost_address): Record inv_expr for all uses of
23732 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23733 (iv_ca_has_deps): Reimplemented to ...
23734 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23736 (iv_ca_extend): Call iv_ca_more_deps.
23738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23740 * tree-ssa-address.c (struct mem_address): Move to header file.
23741 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23742 * tree-ssa-address.h (struct mem_address): Move from C file.
23743 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23745 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23747 * tree-affine.h (aff_combination_type): New interface.
23748 (aff_combination_zero_p): Remove static.
23749 (aff_combination_const_p): New interface.
23750 (aff_combination_singleton_var_p): New interfaces.
23752 2017-05-11 Richard Biener <rguenther@suse.de>
23754 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23755 Skip unreachable blocks and destinations.
23756 (eliminate): Move stmt removal and fixup ...
23757 (fini_eliminate): ... here. Skip inserted exprs.
23758 (pass_pre::execute): Move fini_pre after fini_eliminate.
23759 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23760 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23761 PRE to get rid of dead code that has invalid SSA form and
23762 split critical edges again.
23764 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23766 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23768 2017-05-11 Richard Biener <rguenther@suse.de>
23770 * passes.c (execute_function_todo): Verify loops if they are
23771 said to be up-to-date.
23772 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23773 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23775 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23778 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23779 handle calling assemble_external ourself.
23782 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23783 modes with zero size. Enhance comment.
23785 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23787 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23788 built-ins for vec_xl and vec_xst with short and char pointer
23791 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23793 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23794 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23795 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23796 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23797 (_mm_maskz_min_round_ss): New intrinsics.
23798 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23799 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23800 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23801 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23802 (__builtin_ia32_minss_mask_round): New builtins.
23803 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23804 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23805 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23807 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23808 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23810 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23813 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23815 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23816 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23817 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23818 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23819 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23820 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23821 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23822 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23823 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23824 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23825 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23826 (__builtin_ia32_mulss_mask_round): New builtins.
23827 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23828 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23829 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23831 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23832 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23834 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23837 2017-05-10 Julia Koval <julia.koval@intel.com>
23839 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23840 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23841 (_mm256_setr_m128i): New intrinsics.
23843 2017-05-10 Julia Koval <julia.koval@intel.com>
23845 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23846 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23847 (_mm_maskz_rcp14_ss): New intrinsics.
23848 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23849 (__builtin_ia32_rcp14ss_mask): New builtins.
23850 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23852 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23854 PR tree-optimization/51513
23855 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23856 (assert_unreachable_fallthru_edge_p): Use it.
23857 (group_case_labels_stmt): Likewise.
23858 * tree-cfg.h: Prototype it.
23859 * stmt.c: Include cfghooks.h and tree-cfg.h.
23860 (emit_case_dispatch_table) <gap_label>: New local variable.
23861 Use it to fill dispatch table gaps.
23862 Test for default_label before updating probabilities.
23863 (expand_case) <default_label>: Remove unneeded initialization.
23864 Test for unreachable default case statement and remove its edge.
23865 Set default_label accordingly.
23866 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23868 2017-05-10 Carl Love <cel@us.ibm.com>
23870 * config/rs6000/rs6000-c: Add support for built-in functions
23871 vector signed char vec_neg (vector signed char)
23872 vector signed short int vec_neg (vector short int)
23873 vector signed int vec_neg (vector signed int)
23874 vector signed long long vec_neg (vector signed long long)
23875 vector float vec_neg (vector float)
23876 vector double vec_neg (vector double)
23877 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23879 * config/rs6000/altivec.h: Add define for vec_neg
23880 * doc/extend.texi: Update the built-in documentation for the
23881 new built-in functions.
23883 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23885 PR tree-optimization/77644
23886 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23888 2017-05-10 Nathan Sidwell <nathan@acm.org>
23890 * dumpfile.h (TDI_lang_all): New.
23891 (TDF_KIND): New. Renumber others
23892 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23894 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23896 (get_dump_file_name): Adjust suffix generation.
23897 (dump_enable_all): Use TDF_KIND.
23898 * doc/invoke.texi (-fdump-lang-all): Document.
23900 * dumpfile.h: Tabify.
23902 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23905 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23906 Move member access before delete.
23908 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23910 * tree-inline.c (expand_call_inline): Split block at stmt
23913 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23916 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23917 are now unused after splitting mov{sf,sd}_hardfloat.
23918 (f32_lr2): Likewise.
23919 (f32_lm): Likewise.
23920 (f32_lm2): Likewise.
23921 (f32_li): Likewise.
23922 (f32_li2): Likewise.
23923 (f32_lv): Likewise.
23924 (f32_sr): Likewise.
23925 (f32_sr2): Likewise.
23926 (f32_sm): Likewise.
23927 (f32_sm2): Likewise.
23928 (f32_si): Likewise.
23929 (f32_si2): Likewise.
23930 (f32_sv): Likewise.
23931 (f32_dm): Likewise.
23932 (f32_vsx): Likewise.
23933 (f32_av): Likewise.
23934 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23935 For movsf, order stores so the VSX stores occur before the GPR
23936 store which encourages the register allocator to use a traditional
23937 FPR instead of a GPR. For movsd, order the stores so that the GPR
23938 store comes before the VSX stores to allow the power6 to work.
23939 This is due to the power6 not having a 32-bit integer store
23940 instruction from a FPR.
23941 (movsf_hardfloat): Likewise.
23942 (movsd_hardfloat): Likewise.
23944 2017-05-09 Martin Sebor <msebor@redhat.com>
23946 PR translation/80280
23947 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23950 PR translation/80280
23951 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23952 data member added in r247778.
23953 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23955 2017-05-09 Nathan Sidwell <nathan@acm.org>
23957 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23959 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23962 2017-05-09 Marek Polacek <polacek@redhat.com>
23964 * doc/invoke.texi: Fix typo.
23966 2017-05-09 Richard Biener <rguenther@suse.de>
23968 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23969 (vrp_val_is_min): Likewise.
23970 (set_value_range_to_value): Likewise.
23971 (set_value_range_to_nonnegative): Likewise.
23972 (gimple_assign_nonzero_p): Likewise.
23973 (gimple_stmt_nonzero_p): Likewise.
23974 (vrp_int_const_binop): Likewise. Remove unreachable case.
23975 (adjust_range_with_scev): Adjust comments.
23976 (compare_range_with_value): Likewise.
23977 (extract_range_from_phi_node): Likewise.
23978 (test_for_singularity): Likewise.
23980 2017-05-09 Richard Biener <rguenther@suse.de>
23982 * tree-vrp.c (get_single_symbol): Add assert that we don't
23983 get overflowed constants as invariant part.
23984 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23985 checks. Use wi::cmp instead of recursing for integer constants.
23986 (compare_values): Just ignore whether we assumed undefined
23987 overflow instead of failing the compare.
23988 (extract_range_for_var_from_comparison_expr): Add comment before the
23989 TREE_NO_WARNING sets.
23990 (test_for_singularity): Likewise.
23991 (extract_range_from_comparison): Do not disable optimization
23992 when we assumed undefined overflow.
23993 (extract_range_basic): Remove init of unused var.
23995 2017-05-09 Richard Biener <rguenther@suse.de>
23997 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23998 (extract_range_from_multiplicative_op_1): Adjust.
23999 (extract_range_from_binary_expr_1): Use int_const_binop.
24001 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24004 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24005 rs6000_store_data_bypass_p in seven define_bypass directives and
24006 in several comments.
24007 * config/rs6000/rs6000-protos.h: Add prototype for
24008 rs6000_store_data_bypass_p function.
24009 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24010 function implements slightly different (rs6000-specific) semantics
24011 than store_data_bypass_p, returning false rather than aborting
24012 with assertion error when arguments do not satisfy the
24013 requirements of store data bypass.
24014 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24015 rs6000_store_data_bypass_p.
24017 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24019 * config/xtensa/xtensa-protos.h
24020 (xtensa_initial_elimination_offset): New declaration.
24021 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24022 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24023 macro definition, add case for FRAME_POINTER_REGNUM when
24024 FRAME_GROWS_DOWNWARD.
24025 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24026 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24027 xtensa_initial_elimination_offset.
24029 2017-05-08 Nathan Sidwell <nathan@acm.org>
24031 * doc/invoke.texi: Alphabetize -fdump options.
24033 2017-05-08 Martin Sebor <msebor@redhat.com>
24035 PR translation/80280
24036 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24038 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24040 * target.def (compute_frame_layout): New optional target hook.
24041 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24042 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24043 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24045 * reload1.c (verify_initial_elim_offsets): Likewise.
24046 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24047 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24048 (arm_get_frame_offsets): Split up into this ...
24049 (arm_compute_frame_layout): ... and this function.
24051 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24053 * config/aarch64/constraints.md (Usa): New constraint.
24054 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24056 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24058 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24059 with_multilib_list after it has been checked.
24061 2017-05-08 Richard Biener <rguenther@suse.de>
24063 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24064 (bitmap_set_subtract_values): Likewise.
24066 2017-05-08 Richard Biener <rguenther@suse.de>
24068 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24069 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24071 (gimple_stmt_nonzero_warnv_p): Rename to ...
24072 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24074 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24075 (extract_range_basic): Adjust, do not disable propagation on
24076 strict overflow sensitive simplification.
24077 (vrp_visit_cond_stmt): Likewise.
24079 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24081 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24082 body size unconditionally.
24084 2017-05-07 Jeff Law <law@redhat.com>
24087 2017-05-06 Jeff Law <law@redhat.com>
24088 PR tree-optimization/78496
24089 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24092 PR tree-optimization/78496
24093 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24094 (simplify_stmt_using_ranges): Call it.
24095 (vrp_dom_walker::before_dom_children): Extract equivalences
24096 from an ASSERT_EXPR with an equality comparison against a
24099 2017-05-06 Jeff Law <law@redhat.com>
24101 PR tree-optimization/78496
24102 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24105 PR tree-optimization/78496
24106 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24107 (simplify_stmt_using_ranges): Call it.
24108 (vrp_dom_walker::before_dom_children): Extract equivalences
24109 from an ASSERT_EXPR with an equality comparison against a
24112 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24114 * lra-constraints.c (lra_copy_reg_equiv): New function.
24115 (split_reg): Use it to copy equivalence information from the
24116 original register to the spill register.
24118 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24120 PR rtl-optimization/75964
24121 * simplify-rtx.c (simplify_const_relational_operation): Remove
24122 invalid handling of comparisons of integer ABS.
24124 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24126 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24127 initialize to zero.
24128 (init_regs): Remove declaration.
24129 (function_arg_advance_32): Initialize error_p as boolean variable.
24131 2017-05-05 Nathan Sidwell <nathan@acm.org>
24133 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24134 lines. Use for (;;).
24136 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24138 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24139 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24140 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24141 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24142 VF=2 that require versioning.
24144 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24146 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24149 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24151 * diagnostic.h (diagnostic_override_option_index): Convert from
24152 macro to inline function.
24154 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24156 * diagnostic.c (last_module_changed_p): New function.
24157 (set_last_module): New function.
24158 (diagnostic_report_current_module): Convert macro usage to
24159 the above functions.
24160 * diagnostic.h (diagnostic_context::last_module): Strengthen
24161 from const line_map * to const line_map_ordinary *.
24162 (diagnostic_last_module_changed): Delete macro.
24163 (diagnostic_set_last_module): Delete macro.
24165 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24167 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24168 with diagnostic_report_diagnostic.
24169 (diagnostic_n_impl_richloc): Likewise.
24170 * diagnostic.h (report_diagnostic): Delete macro.
24171 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24172 with diagnostic_report_diagnostic.
24173 * substring-locations.c (format_warning_va): Likewise.
24175 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24177 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24178 save/restor of format_spec. Move option-printing code to...
24179 (print_option_information): ...this new function, and
24180 reimplement by simply printing to the pretty_printer,
24181 rather than appending to the format string.
24183 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24185 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24186 handling logic into...
24187 (update_effective_level_from_pragmas): ...this new function.
24189 2017-05-04 Andrew Waterman <andrew@sifive.com>
24191 * config/riscv/riscv.opt (mstrict-align): New option.
24192 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24193 (SLOW_UNALIGNED_ACCESS): Define.
24194 (riscv_slow_unaligned_access): Declare.
24195 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24197 (riscv_slow_unaligned_access): New variable.
24198 (rocket_tune_info): Set slow_unaligned_access to true.
24199 (optimize_size_tune_info): Set slow_unaligned_access to false.
24200 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24201 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24202 (riscv_option_override): Set riscv_slow_unaligned_access.
24203 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24205 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24207 * config/riscv/riscv.md: Unify indentation.
24209 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24214 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24216 (extendsi<mode>2): Add support for doing sign extension via
24217 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24218 don't have ISA 3.0 instructions.
24219 (extendsi<mode>2 splitter): Likewise.
24220 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24221 generate the normal insns since SImode can now go in vector
24222 registers. Disallow the special UNSPECs needed for previous
24223 machines to hide SImode being used. Add new insns
24224 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24225 (fix_trunc<mode>si2_stfiwx): Likewise.
24226 (fix_trunc<mode>si2_internal): Likewise.
24227 (fixuns_trunc<mode>si2): Likewise.
24228 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24229 (fctiw<u>z_<mode>_smallint): Likewise.
24230 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24231 of floating point to 32-bit integer from doing a direct move to
24232 the GPR registers to do a store.
24233 (fctiwz_<mode>): Break long line.
24235 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24237 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24238 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24239 (addr_list, addr_offset_valid_p): New.
24240 (split_address_groups): Check offset validity with above function.
24241 (gt-tree-ssa-loop-ivopts.h): Include header file.
24243 2017-05-05 Nathan Sidwell <nathan@acm.org>
24245 * config.gcc (arm*-*-*): Add missing 'fi'.
24247 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24249 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24250 included in -fopt-info does not matter.
24251 * doc/optinfo.texi (-fopt-info): Fix description of default
24252 behavour. Explicitly say order of options included in -fopt-info
24255 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24257 * config.gcc: Allow combinations of aprofile and rmprofile values for
24258 --with-multilib-list.
24259 * config/arm/t-multilib: New file.
24260 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24261 variables. Remove setting of ISA and floating-point ABI in
24262 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24263 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24264 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24265 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24267 * config/arm/t-rmprofile: Likewise except for the matches changes.
24268 * doc/install.texi (--with-multilib-list): Document the combination of
24269 aprofile and rmprofile values and warn about pitfalls in doing that.
24271 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24273 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24274 (movdi_aarch64): Likewise.
24276 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24278 PR tree-optimization/80632
24279 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24281 (build_arrays): Initialize it for virtual phis.
24282 (fix_phi_nodes): Use it for virtual phis.
24284 PR tree-optimization/80558
24285 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24286 [x, y] op z into [x op, y op z] for op & or | if conditions
24289 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24290 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24293 * config/arm/arm.md (use_literal_pool): Remove.
24294 (64-bit immediate split): No longer takes cost into consideration
24295 if arm_disable_literal_pool is enabled.
24296 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24297 used when arm_disable_literal_pool is enabled.
24298 (arm_max_const_double_inline_cost): Remove use of
24299 arm_disable_literal_pool.
24300 (push_minipool_fix): Add assert.
24301 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24302 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24303 (no_literal_pool_sf_immediate): New.
24305 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24307 PR tree-optimization/80613
24308 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24309 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24311 2017-05-05 Richard Biener <rguenther@suse.de>
24313 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24315 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24317 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24318 of this flag from insn conditions due to removal from r247495.
24320 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24322 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24324 (arm_early_store_addr_dep_ptr): Likewise.
24325 * config/arm/aarch-common-protos.h
24326 (arm_early_load_addr_dep_ptr): Add prototype.
24327 (arm_early_store_addr_dep_ptr): Likewise.
24328 * config/arm/cortex-a53.md: Add new bypasses.
24330 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24332 * tree.c (next_type_uid): Change type to unsigned.
24333 (type_hash_canon): Decrement back next_type_uid if
24334 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24335 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24338 2017-05-04 Martin Sebor <msebor@redhat.com>
24340 * builtins.c: Fix a trivial typo in a comment.
24342 PR middle-end/79234
24343 * builtins.c (check_sizes): Adjust to handle reading past the end.
24344 Avoid printing excessive upper bound of ranges. Use %E to print
24345 tree nodes instead of converting them to %wu.
24346 (expand_builtin_memchr): New function.
24347 (compute_dest_size): Rename...
24348 (compute_objsize): ...to this.
24349 (expand_builtin_memcpy): Adjust.
24350 (expand_builtin_mempcpy): Adjust.
24351 (expand_builtin_strcat): Adjust.
24352 (expand_builtin_strcpy): Adjust.
24353 (check_strncat_sizes): Adjust.
24354 (expand_builtin_strncat): Adjust.
24355 (expand_builtin_strncpy): Adjust and simplify.
24356 (expand_builtin_memset): Adjust.
24357 (expand_builtin_bzero): Adjust.
24358 (expand_builtin_memcmp): Adjust.
24359 (expand_builtin): Handle memcmp.
24360 (maybe_emit_chk_warning): Check strncat just once.
24362 2017-05-04 Martin Sebor <msebor@redhat.com>
24364 PR preprocessor/79214
24365 PR middle-end/79222
24366 PR middle-end/79223
24367 * builtins.c (check_sizes): Add inlining context and issue
24368 warnings even when -Wno-system-headers is set.
24369 (check_strncat_sizes): Same.
24370 (expand_builtin_strncat): Same.
24371 (expand_builtin_memmove): New function.
24372 (expand_builtin_stpncpy): Same.
24373 (expand_builtin): Handle memmove and stpncpy.
24375 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24377 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24378 which is not used any more.
24380 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24382 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24384 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24386 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24387 (cortexa53_tunings): Likewise.
24388 (cortexa57_tunings): Likewise.
24389 (cortexa72_tunings): Likewise.
24390 (cortexa73_tunings): Likewise.
24392 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24394 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24395 Set loop alignment to 8.
24397 2017-05-04 Martin Sebor <msebor@redhat.com>
24399 PR translation/80280
24400 * builtins.c (expand_builtin_object_size): Add missing quoting to
24401 %D and like directives.
24402 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24403 (hsa_type_for_tree_type): Same.
24404 (verify_function_arguments): Same.
24405 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24406 * varasm.c (get_section): Same.
24409 2017-05-04 Martin Sebor <msebor@redhat.com>
24411 PR translation/80280
24412 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24414 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24416 * config/aarch64/aarch64.c (generic_addrcost_table):
24417 Change HI/TI mode setting.
24419 2017-05-04 Martin Jambor <mjambor@suse.cz>
24421 PR tree-optimization/80622
24422 * tree-sra.c (comes_initialized_p): New function.
24423 (build_accesses_from_assign): Only set write lazily when
24424 comes_initialized_p is false.
24425 (analyze_access_subtree): Use comes_initialized_p.
24426 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24427 instead of testing for PARM_DECL.
24429 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24431 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24432 constraint on operand 0 to allow more general addressing modes.
24433 Adjust output template.
24434 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24436 * config/aarch64/aarch64-protos.h
24437 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24438 * config/aarch64/constraints.md (Dp): New address constraint.
24439 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24442 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24444 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24445 update use of estimate_ipcp_clone_size_and_time.
24446 (estimate_local_effects): Update use of
24447 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24448 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24449 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24450 Return nonspecialized time.
24452 2017-05-04 Richard Biener <rguenther@suse.de>
24454 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24455 for the last VUSE which def dominates the PHI. Directly call
24457 (get_continuation_for_phi_1): Remove.
24459 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24461 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24462 to explain the use of truncating division. Cap the number of
24463 iterations to the maximum given by nb_iterations_upper_bound,
24466 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24468 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24469 * configure: Regenerate.
24470 * config.in: Regenerate.
24471 * config/i386/driver-mingw32.c: new file.
24472 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24473 * config.host: Link driver-mingw32.o on MinGW host.
24474 * doc/install.texi: Document new --enable-mingw-wildcard configure
24477 2017-05-04 Marek Polacek <polacek@redhat.com>
24479 PR tree-optimization/80612
24480 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24482 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24483 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24485 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24486 (movt splitter): Likewise.
24487 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24488 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24489 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24490 block for Thumb-1 with MOVT.
24491 (thumb2_legitimate_address_p): Move code block ...
24492 (can_avoid_literal_pool_for_label_p): ... into this new function.
24493 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24495 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24496 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24497 "M-profile targets with the MOVT instruction".
24499 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24501 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24502 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24503 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24505 2017-05-04 Martin Liska <mliska@suse.cz>
24507 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24508 variable cond_code.
24510 2017-05-04 Richard Biener <rguenther@suse.de>
24512 * tree.c (array_at_struct_end_p): Handle arrays at struct
24513 end with flexarrays more conservatively. Refactor and treat
24514 arrays of arrays or aggregates more strict. Fix
24515 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24516 * tree.c (array_at_struct_end_p): Adjust prototype.
24517 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24518 * gimple-fold.c (get_range_strlen): Likewise.
24519 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24521 2017-05-04 Richard Biener <rguenther@suse.de>
24523 PR tree-optimization/31130
24524 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24526 (supports_overflow_infinity): Likewise.
24527 (is_negative_overflow_infinity): Likewise.
24528 (is_positive_overflow_infinity): Likewise.
24529 (is_overflow_infinity): Likewise.
24530 (stmt_overflow_infinity): Likewise.
24531 (overflow_infinity_range_p): Likewise.
24532 (usable_range_p): Remove as always returning true.
24533 (make_overflow_infinity): Remove.
24534 (negative_overflow_infinity): Likewise.
24535 (positive_overflow_infinity): Likewise.
24536 (avoid_overflow_infinity): Likewise.
24537 (set_value_range): Adjust accordingly.
24538 (set_value_range_to_nonnegative): Likewise, remove now unused
24539 overflow_infinity arg.
24540 (vrp_operand_equal_p): Adjust.
24541 (update_value_range): Likewise.
24542 (range_int_cst_singleton_p): Likewise.
24543 (operand_less_p): Likewise.
24544 (compare_values_warnv): Likewise.
24545 (extract_range_for_var_from_comparison_expr): Likewise.
24546 (vrp_int_const_binop): Likewise.
24547 (zero_nonzero_bits_from_vr): Likewise.
24548 (extract_range_from_multiplicative_op_1): Likewise.
24549 (extract_range_from_binary_expr_1): Likewise.
24550 (extract_range_from_unary_expr): Likewise.
24551 (extract_range_from_comparison): Likewise.
24552 (extract_range_basic): Likewise.
24553 (adjust_range_with_scev): Likewise.
24554 (compare_ranges): Likewise.
24555 (compare_range_with_value): Likewise.
24556 (dump_value_range): Likewise.
24557 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24559 (simplify_cond_using_ranges): Adjust.
24561 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24563 * brig-builtins.def: Added a builtin for class_f64.
24564 * builtin-types.def: Added a builtin type needed by class_f64.
24566 2017-05-03 Jason Merrill <jason@redhat.com>
24568 * timevar.def: Add TV_CONSTEXPR.
24570 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24572 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24574 2017-05-03 Martin Jambor <mjambor@suse.cz>
24576 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24577 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24578 * ipa-cp.c (ipcp_propagate_stage): Do not call
24579 ipa_update_after_lto_read.
24580 * ipa-inline.c (ipa_inline): Likewise.
24582 2017-05-03 Martin Jambor <mjambor@suse.cz>
24584 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24585 tag. Added a default constructor and a destructor.
24586 (ipa_edge_args_sum_t): New class;
24587 (ipa_edge_args_sum): Declare.
24588 (ipa_edge_args_vector): Remove declaration.
24589 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24590 (ipa_free_edge_args_substructures): Remove declaration.
24591 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24592 (ipa_edge_args_info_available_for_edge_p): Likewise.
24593 * ipa-prop.c (ipa_edge_args_vector): Removed.
24594 (edge_removal_hook_holder): Likewise.
24595 (edge_duplication_hook_holder): Likewise.
24596 (ipa_edge_args_sum): New variable.
24597 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24598 ipa_edge_args_vector.
24599 (ipa_free_edge_args_substructures): Likewise.
24600 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24601 ipa_edge_args_vector.
24602 (ipa_edge_removal_hook): Turned into method
24603 ipa_edge_args_sum_t::remove.
24604 (ipa_edge_duplication_hook): Turned into method
24605 ipa_edge_args_sum_t::duplicate.
24606 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24607 registering edge hooks.
24608 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24609 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24610 ipa_edge_args_sum instead of ipa_edge_args_vector.
24611 * ipa-profile.c (ipa_profile): Likewise.
24613 2017-05-03 Martin Jambor <mjambor@suse.cz>
24615 * symbol-summary.h (function_summary): New method exists.
24616 (function_summary::symtab_removal): Deallocate through release.
24617 (call_summary): New class.
24618 (gt_ggc_mx): New overload.
24619 (gt_pch_nx): Likewise.
24620 (gt_pch_nx): Likewise.
24622 2017-05-03 Jeff Law <law@redhat.com>
24624 PR tree-optimization/78496
24625 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24626 from simplify_cond_using_ranges. Split off code to walk
24627 backwards through casts into ...
24628 (simplify_cond_using_ranges_2): New function.
24629 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24630 (execute_vrp): After identifying jump threads, call
24631 simplify_cond_using_ranges_2.
24633 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24636 * ipa-inline.h (inline_summary): Add ctor.
24637 (create_ggc): Do not use ggc_cleared_alloc.
24639 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24640 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24642 * gcc.c (handle_braces): Support escaping in switch matching
24644 * doc/invoke.texi (Spec Files): Document it.
24645 Remove superfluous @code markup in items.
24647 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24649 * diagnostic-show-locus.c (struct column_range): New struct.
24650 (get_affected_columns): New function.
24651 (get_printed_columns): New function.
24652 (struct correction): New struct.
24653 (correction::ensure_capacity): New function.
24654 (correction::ensure_terminated): New function.
24655 (struct line_corrections): New struct.
24656 (line_corrections::~line_corrections): New dtor.
24657 (line_corrections::add_hint): New function.
24658 (layout::print_trailing_fixits): Reimplement in terms of the new
24660 (selftest::test_overlapped_fixit_printing): New function.
24661 (selftest::diagnostic_show_locus_c_tests): Call it.
24663 2017-05-03 Nathan Sidwell <nathan@acm.org>
24665 Canonicalize canonical type hashing
24666 * tree.h (type_hash_canon_hash): Declare.
24667 * tree.c (type_hash_list, attribute_hash_list): Move into
24668 type_hash_canon_hash.
24669 (build_type_attribute_qual_variant): Break out hash code calc into
24670 type_hash_canon_hash.
24671 (type_hash_canon_hash): New. Generic type hash computation.
24672 (build_range_type_1, build_array_type_1, build_function_type,
24673 build_method_type_directly, build_offset_type, build_complex_type,
24674 make_vector_type): Call it.
24676 2017-05-03 Richard Biener <rguenther@suse.de>
24678 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24679 When all DRs have unknown misaligned do not always peel
24680 when there is a store but apply the same costing model as if
24681 there were only loads.
24683 2017-05-03 Richard Biener <rguenther@suse.de>
24686 PR tree-optimization/80492
24687 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24688 compare_base_decls returning dont-know properly.
24690 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24692 * config/arm/iterators.md (CCSI): New mode iterator.
24693 (arch): New mode attribute.
24694 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24695 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24696 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24697 code iterator for success result mode.
24698 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24699 the corresponding new insn generators.
24701 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24704 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24705 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24707 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24709 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24710 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24711 (DDR_A): Wrap DDR argument in brackets.
24712 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24713 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24714 (DDR_REVERSED_P): Likewise.
24716 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24718 PR tree-optimization/79472
24719 * tree-switch-conversion.c (struct switch_conv_info): Add
24720 contiguous_range and default_case_nonstandard fields.
24721 (collect_switch_conv_info): Compute contiguous_range and
24722 default_case_nonstandard fields, don't clear final_bb if
24723 contiguous_range and only the default case doesn't have the required
24725 (check_all_empty_except_final): Set default_case_nonstandard instead
24726 of failing if contiguous_range and the default case doesn't have empty
24728 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24729 and only the default case doesn't have the required constants. Skip
24731 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24732 if default_case_nonstandard.
24733 (build_constructors): Build constant 1 just once. Assert that default
24734 values aren't inserted in between cases if contiguous_range. Skip
24736 (build_arrays): Skip virtual phis.
24737 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24738 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24739 Handle virtual phis.
24740 (gen_inbound_check): Handle default_case_nonstandard case.
24741 (process_switch): Adjust check_final_bb caller. Call
24742 gather_default_values with the first non-default case instead of
24743 default case if default_case_nonstandard.
24745 2017-05-02 Nathan Sidwell <nathan@acm.org>
24747 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24748 check. Fix formatting.
24750 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24752 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24753 errors when comparing specialized and unspecialized times.
24755 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24757 * diagnostic-show-locus.c
24758 (layout::should_print_annotation_line_p): Make private.
24759 (layout::print_annotation_line): Make private.
24760 (layout::annotation_line_showed_range_p): Make private.
24761 (layout::show_ruler): Make private.
24762 (layout::print_source_line): Make private. Pass in line and
24763 line_width, rather than calling location_get_source_line. Drop
24765 (layout::print_leading_fixits): New method.
24766 (layout::print_any_fixits): Rename to...
24767 (layout::print_trailing_fixits): ...this, and make private.
24768 Don't print newline fixits.
24769 (diagnostic_show_locus): Move logic for printing one row into...
24770 (layout::print_line): ...this new function. Move the
24771 location_get_source_line call and error-handling from
24772 print_source_line to here. Call print_leading_fixits, and rename
24773 print_any_fixits to print_trailing_fixits.
24774 (selftest::test_fixit_insert_containing_newline): Update now that
24775 newlines are partially supported.
24776 (selftest::test_fixit_insert_containing_newline_2): New test.
24777 (selftest::test_fixit_replace_containing_newline): Update comments.
24778 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24779 * edit-context.c (class added_line): New class.
24780 (class edited_line): Describe newline handling in comment.
24781 (edited_line::actually_edited_p): New method.
24782 (edited_line::print_content): Delete redundant decl.
24783 (edited_line::m_predecessors): New field.
24784 (edited_file::print_content): Call edited_line::print_content.
24785 (edited_file::print_diff): Update to support newlines.
24786 (edited_file::print_diff_hunk): Likewise.
24787 (edited_file::print_run_of_changed_lines): New function.
24788 (edited_file::print_diff_line): Convert to...
24789 (print_diff_line): ...this.
24790 (edited_file::get_effective_line_count): New function.
24791 (edited_line::edited_line): Initialize new field m_predecessors.
24792 (edited_line::~edited_line): Clean up m_predecessors.
24793 (edited_line::apply_fixit): Handle newlines.
24794 (edited_line::get_effective_line_count): New function.
24795 (edited_line::print_content): New function.
24796 (edited_line::print_diff_lines): New function.
24797 (selftest::test_applying_fixits_insert_containing_newline): New
24799 (selftest::test_applying_fixits_replace_containing_newline): New
24801 (selftest::insert_line): New function.
24802 (selftest::test_applying_fixits_multiple_lines): Add example of
24804 (selftest::edit_context_c_tests): Call the new tests.
24806 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24808 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24809 parameter cand. Update dump information.
24810 (get_computation_cost): Update uses.
24812 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24814 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24815 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24816 (get_computation_at, rewrite_use_address): Update use of
24817 get_computation_aff.
24819 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24821 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24822 (get_computation): Delete.
24823 (get_computation_cost): Implement like get_computation_cost_at.
24824 Use get_computation_at.
24825 (get_computation_cost_at): Delete.
24826 (rewrite_use_nonlinear_expr): Use get_computation_at.
24827 (rewrite_use_compare, remove_unused_ivs): Ditto.
24829 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24831 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24833 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24835 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24836 (ivopts_global_cost_for_size): Rename parameter and update uses.
24837 (iv_ca_recount_cost): Update uses.
24838 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24839 candidates seperately in n_invs and n_cands.
24840 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24842 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24844 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24845 (find_inv_vars_cb): New.
24846 (find_depends): Renamed to ...
24847 (find_inv_vars): ... this.
24848 (add_candidate_1, force_var_cost): Call find_inv_vars.
24849 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24851 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24853 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24854 inv_vars. Add inv_exprs.
24855 (struct iv_cand): Rename depends_on to inv_vars.
24856 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24857 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24858 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24859 (dump_cand): Dump inv_vars.
24860 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24861 (record_invariant, find_depends, add_candidate_1): Ditto.
24862 (set_group_iv_cost, force_var_cost): Ditto.
24863 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24864 (get_computation_cost_at, get_computation_cost): Ditto.
24865 (determine_group_iv_cost_generic): Ditto.
24866 (determine_group_iv_cost_address): Ditto.
24867 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24868 (determine_group_iv_costs): Ditto.
24869 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24870 (iv_ca_set_remove_invariants): Renamed to ...
24871 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24872 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24873 (iv_ca_set_add_invariants): Renamed to ...
24874 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24875 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24876 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24877 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24878 (create_new_ivs): Remove useless dump.
24880 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24882 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24884 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24885 (iv_ca_set_no_cp, create_new_iv): Ditto.
24887 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24889 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24891 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24893 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24894 function tree_check2.
24896 2017-05-02 Martin Liska <mliska@suse.cz>
24898 * doc/gcov.texi: Add missing preposition.
24899 * gcov.c (function_info::function_info): Properly fill up
24900 all member variables.
24902 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24904 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24906 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24908 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24910 2017-05-02 Martin Liska <mliska@suse.cz>
24913 * lto-streamer-in.c (lto_read_tree_1): Remove
24914 LTO_STREAMER_DEBUG.
24915 * lto-streamer.c (struct tree_hash_entry): Likewise.
24916 (struct tree_entry_hasher): Likewise.
24917 (tree_entry_hasher::hash): Likewise.
24918 (tree_entry_hasher::equal): Likewise.
24919 (lto_streamer_init): Likewise.
24920 (lto_orig_address_map): Likewise.
24921 (lto_orig_address_get): Likewise.
24922 (lto_orig_address_remove): Likewise.
24923 * lto-streamer.h: Likewise.
24924 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24925 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24927 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24929 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24930 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24931 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24932 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24933 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24934 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24935 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24936 (mm_maskz_sub_ss): New intrinsics.
24937 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24938 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24939 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24940 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24941 (__builtin_ia32_subss_mask_round): New builtins.
24942 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24943 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24944 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24946 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24947 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24949 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24952 2017-05-02 Martin Jambor <mjambor@suse.cz>
24954 PR tree-optimization/78687
24955 * tree-sra.c (access): New field parent.
24956 (process_subtree_disqualification): New function.
24957 (disqualify_candidate): Call it.
24958 (build_accesses_from_assign): Reset write flag if creating an
24960 (build_access_subtree): Fill in parent field and also prpagate
24961 down grp_write flag.
24962 (create_artificial_child_access): New parameter set_grp_write, set
24963 grp_write to its value.
24964 (propagate_subaccesses_across_link): Also propagate grp_write flag
24966 (propagate_all_subaccesses): Push the closest parent back to work
24967 queue if add_access_to_work_queue returned true.
24969 2017-05-02 Richard Biener <rguenther@suse.de>
24971 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24972 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24973 -fstrict-overflow documentation.
24974 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24975 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24976 flag_strict_overflow.
24977 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24978 * lto-opts.c (lto_write_options): Do not stream it.
24979 * lto-wrapper.c (merge_and_complain): Do not handle it.
24980 * opts.c (default_options_table): Do not set -fstrict-overflow.
24981 (finish_options): Likewise do not clear it when sanitizing.
24982 * simplify-rtx.c (simplify_const_relational_operation): Do not
24983 test flag_strict_overflow.
24985 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24987 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24988 using enabled attribute.
24989 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24990 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24991 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24992 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24993 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24994 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24995 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24996 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24997 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24998 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25000 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25002 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25004 2017-05-02 Richard Biener <rguenther@suse.de>
25006 PR tree-optimization/80591
25008 2017-04-10 Richard Biener <rguenther@suse.de>
25010 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25013 2017-05-02 Richard Biener <rguenther@suse.de>
25015 PR tree-optimization/80549
25016 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25017 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25018 headers if they do not have a preheader.
25020 2017-05-02 Martin Liska <mliska@suse.cz>
25023 * common.opt: Fix typo.
25024 * doc/invoke.texi: Likewise.
25026 2017-05-01 Jan Beulich <jbeulich@suse.com>
25028 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25029 swapping, add (x,x,m,x,n) alternative.
25031 2017-05-01 Nathan Sidwell <nathan@acm.org>
25033 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25034 unnecessary unadjusted_alignment check.
25036 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25039 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25041 * gimplify.c (gimplify_cilk_detach): New function.
25042 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25043 * tree-core.h: Document EXPR_CILK_SPAWN.
25044 * tree.h (EXPR_CILK_SPAWN): Define.
25046 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25048 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25049 to use new fixit_hint representation, using the "replace" logic.
25050 (get_line_span_for_fixit_hint): Likewise.
25051 (layout::print_any_fixits): Likewise.
25052 (selftest::test_one_liner_many_fixits): Rename to...
25053 (selftest::test_one_liner_many_fixits_1): ...this, and update
25054 comment and expected output to reflect that the multiple fix-it
25055 hints are now consolidated into one insertion.
25056 (selftest::test_one_liner_many_fixits_2): New test.
25057 (selftest::test_diagnostic_show_locus_one_liner): Update for
25059 (selftest::test_fixit_consolidation): Update for fix-it API
25061 * diagnostic.c (print_parseable_fixits): Likewise.
25062 * edit-context.c (edited_line::m_line_events): Convert from
25063 auto_vec <line_event *> to auto_vec <line_event>.
25064 (class line_event): Convert from abstract base class to a concrete
25065 class, taking over the role of replace_event.
25066 (class insert_event): Delete.
25067 (class replace_event): Rename to class line_event. Convert to
25069 (edit_context::add_fixits): Reimplement.
25070 (edit_context::apply_insert): Delete.
25071 (edit_context::apply_replace): Rename to...
25072 (edit_context::apply_fixit): ...this. Convert to half-open range.
25073 (edited_file::apply_insert): Delete.
25074 (edited_file::apply_replace): Rename to...
25075 (edited_file::apply_fixit): ...this.
25076 (edited_line::~edited_line): Drop deletion of events.
25077 (edited_line::apply_insert): Delete.
25078 (edited_line::apply_replace): Rename to...
25079 (edited_line::apply_fixit): ...this. Convert to half-open range.
25080 Update for change to type of m_line_events.
25081 * edit-context.h (edit_context::apply_insert): Delete.
25082 (edit_context::apply_replace): Rename to...
25083 (edit_context::apply_fixit): ...this.
25085 2017-05-01 Martin Sebor <msebor@redhat.com>
25087 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25090 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25093 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25094 __get_cpuid_max returns 0.
25095 (__get_cpuid_count): Ditto.
25097 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25099 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25100 replacement expression is another instance of one of its arguments.
25102 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25105 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25106 check for stack push/pop autoinc.
25107 * config/i386/i386.c (ix86_agi_dependent): Return false
25108 if the only reason why modified_in_p returned true is that
25109 addr is SP based and set_insn is a push or pop.
25111 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25113 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25116 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25119 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25120 (account_size_time): Use two predicates - exec_pred and
25122 (evaluate_conditions_for_known_args): Compute both clause and
25124 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25125 (inline_summary_t::duplicate): Update.
25126 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25128 (compute_inline_parameters): Likewise.
25129 (estimate_edge_size_and_time): Update caluclation of time.
25130 (estimate_node_size_and_time): Compute both time and nonspecialized
25132 (estimate_ipcp_clone_size_and_time): Update.
25133 (inline_merge_summary): Update.
25134 (do_estimate_edge_time): Update.
25135 (do_estimate_edge_size): Update.
25136 (do_estimate_edge_hints): Update.
25137 (inline_read_section, inline_write_summary): Stream both new predicates.
25138 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25140 (compute_inlined_call_time): Cleanup.
25141 (big_speedup_p): Update.
25142 (edge_badness): Update.
25143 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25144 (size_time_entry): Replace predicate by exec_predicate and
25145 nonconst_predicate.
25146 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25147 (estimate_edge_time): Return also nonspec_time.
25148 (reset_edge_growth_cache): Update.
25150 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25152 PR rtl-optimization/80491
25153 * ifcvt.c (noce_process_if_block): When looking for x setter
25154 with missing else_bb, don't check only the insn right before
25155 cond_earliest, but look for the last insn that x is modified in
25156 within the same bb.
25158 PR rtl-optimization/80491
25159 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25161 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25163 PR tree-optimization/80487
25164 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25166 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25168 PR tree-optimization/79697
25169 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25170 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25171 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25173 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25174 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25176 2017-04-28 Martin Sebor <msebor@redhat.com>
25178 PR tree-optimization/80523
25179 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25180 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25182 (maybe_warn, format_directive, parse_directive): Use new functions.
25183 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25185 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25187 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25189 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25191 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25192 target_header_dir): Set correctly.
25193 * configure: Regenerated.
25194 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25195 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25196 instead of SYSTEM_HEADER_DIR.
25198 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25200 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25201 (estimate_local_effects): Likewise.
25202 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25203 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25204 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25205 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25206 do_estimate_edge_time, estimate_edge_time): Likewise.
25207 * ipa-inline-analysis.c (estimate_node_size_and_time,
25208 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25209 (estimate_time_after_inlining): Remove.
25211 2017-04-28 Martin Liska <mliska@suse.cz>
25213 * doc/gcov.texi: Enhance documentation of gcov.
25215 2017-04-28 Martin Liska <mliska@suse.cz>
25217 * doc/gcov.texi: Sort options in alphabetic order.
25218 * doc/gcov-dump.texi: Likewise.
25219 * doc/gcov-tool.texi: Likewise.
25220 * gcov.c (print_usage): Likewise.
25221 * gcov-dump.c (print_usage): Likewise.
25222 * gcov-tool.c (print_merge_usage_message): Likewise.
25223 (print_rewrite_usage_message): Likewise.
25224 (print_overlap_usage_message): Likewise.
25226 2017-04-28 Martin Liska <mliska@suse.cz>
25228 PR gcov-profile/53915
25229 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25231 2017-04-28 Martin Liska <mliska@suse.cz>
25233 PR gcov-profile/79891
25234 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25235 is marked by compiler as living on a line.
25236 (get_cycles_count): Remove usage of the union.
25237 (output_intermediate_file): Likewise.
25238 (find_source): Fix GNU coding style.
25239 (accumulate_line_counts): Remove old non-all block mode.
25240 (output_lines): Remove usage of the union.
25241 * profile.c (output_location): Include all BBs, even if
25242 belonging to a same line (and file) as a previous BB.
25244 2017-04-28 Martin Liska <mliska@suse.cz>
25246 * gcov.c (process_args): Handle new argument 'w'.
25247 (read_graph_file): Assign ID to BBs.
25248 (output_branch_count): Display BB # if verbose flag is set.
25249 (output_lines): Likewise for arcs.
25250 (print_usage): Add '--verbose' option help.
25251 * doc/gcov.texi: Document --verbose (-w) option.
25253 2017-04-28 Martin Liska <mliska@suse.cz>
25255 * gcov.c (struct block_location_info): New struct.
25256 (process_file): Fill up the new structure.
25257 (read_graph_file): Replace usage of encoding by the newly added
25259 (add_line_counts): Likewise.
25260 (accumulate_line_counts): Remove usage of the union.
25261 (function_info::function_info): New function.
25262 (function_info::~function_info): Likewise.
25263 (process_file): Call delete instead of release_function.
25264 (release_function): Release the function.
25265 (release_structures): Call delete instead of release_function.
25266 (solve_flow_graph): Replace usage of num_blocks.
25267 (find_exception_blocks): Likewise.
25268 (output_lines): Fix GNU coding style.
25270 2017-04-28 Martin Liska <mliska@suse.cz>
25273 * coverage.c (coverage_remove_note_file): New function.
25274 * coverage.h: Declare the function.
25275 * toplev.c (finalize): Clean if an error has been seen.
25277 2017-04-28 Martin Liska <mliska@suse.cz>
25279 PR gcov-profile/80031
25280 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25281 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25282 * gcov.c (read_graph_file): Read just number of blocks.
25283 * profile.c (branch_prob): Do not stream 0 flags per a basic
25286 2017-04-28 Martin Liska <mliska@suse.cz>
25288 * gcov-dump.c (tag_*): Add new argument to declarations.
25289 (dump_gcov_file): Likewise.
25290 (tag_blocks): Add and use new argument depth.
25291 (tag_arcs): Likewise.
25292 (tag_lines): Likewise.
25293 (tag_counters): Likewise.
25294 (tag_summary): Likewise.
25295 (dump_working_sets): Use depth to do a proper indentation.
25297 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25300 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25301 * symtab.c (symtab_node::debug_symtab): Move definition here.
25303 2017-04-28 Richard Biener <rguenther@suse.de>
25305 * lto-streamer.h (LTO_major_version): Bump to 7.
25307 2017-04-28 Richard Biener <rguenther@suse.de>
25309 * tree-vrp.c (assert_info): New struct.
25310 (add_assert_info): New helper.
25311 (register_edge_assert_for_2): Refactor to add asserts to a vector
25313 (register_edge_assert_for_1): Likewise.
25314 (register_edge_assert_for): Likewise.
25315 (finish_register_edge_assert_for): New helper actually registering
25316 asserts where live on edge.
25317 (find_conditional_asserts): Adjust.
25318 (find_switch_asserts): Likewise.
25319 (evrp_dom_walker::try_find_new_range): Generalize.
25320 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25322 2017-04-27 Marek Polacek <polacek@redhat.com>
25325 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25326 arg10 and arg11 to itype.
25328 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25330 * doc/extend.texi (Object Size Checking): Improve grammar.
25332 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25335 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25336 that the logic for permitting reciprocal estimates matches that
25339 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25342 * tree.c (type_cache_hasher::equal): Only compare
25343 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25344 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25345 non-aggregate element types.
25346 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25347 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25349 2017-04-27 Richard Biener <rguenther@suse.de>
25351 PR middle-end/80533
25352 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25353 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25354 keeping a reference to a trailing array.
25356 2017-04-27 Richard Biener <rguenther@suse.de>
25358 PR middle-end/80539
25359 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25360 being in loop-closed SSA form conservatively.
25361 (chrec_fold_multiply_poly_poly): Likewise.
25363 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25365 PR middle-end/79665
25366 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25367 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25369 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25372 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25373 (aarch64_function_arg_alignment): Return unsigned int again, but still
25374 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25375 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25376 Don't emit -Wpsabi note.
25377 (aarch64_function_arg_boundary): Likewise.
25378 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25381 2017-04-26 Nathan Sidwell <nathan@acm.org>
25383 * tree.h (crc32_unsigned_n): Declare.
25384 (crc32_unsigned, crc32_unsigned): Make inline.
25385 * tree.c (crc32_unsigned_bits): Replace with ...
25386 (crc32_unsigned_n): ... this.
25387 (crc32_unsigned, crc32_byte): Remove.
25388 (crc32_string): Remove unnecessary braces.
25390 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25392 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25393 * ipa-inline-analysis.c (MAX_TIME): Remove.
25394 (account_size_time): Use sreal for time.
25395 (dump_inline_summary): Update.
25396 (estimate_function_body_sizes): Update.
25397 (estimate_edge_size_and_time): Update.
25398 (estimate_calls_size_and_time): Update.
25399 (estimate_node_size_and_time): Update.
25400 (inline_merge_summary): Update.
25401 (inline_update_overall_summary): Update.
25402 (estimate_time_after_inlining): Update.
25403 (inline_read_section): Update.
25404 (inline_write_summary): Update.
25405 * ipa-inline.c (compute_uninlined_call_time): Update.
25406 (compute_inlined_call_time): Update.
25407 (recursive_inlining): Update.
25408 (inline_small_functions): Update.
25409 (dump_overall_stats): Update.
25410 * ipa-inline.h: Include sreal.h.
25411 (size_time_entry): Turn time to sreal.
25412 (inline_summary): Turn self_time nad time to sreal.
25414 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25416 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25418 (sreal::stream_out, sreal::stream_in): New.
25419 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25421 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25423 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25426 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25429 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25431 (dimode_scalar_chain::compute_convert_gain): Ditto.
25432 (dimode_scalar_chain::make_vector_copies): Ditto.
25433 (dimode_scalar_chain::convert_reg): Ditto.
25434 (dimode_scalar_chain::convert_insn): Ditto.
25435 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25436 (VI248_AVX512BW_1): New mode iterator.
25437 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25438 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25441 2017-04-25 Martin Sebor <msebor@redhat.com>
25443 PR tree-optimization/80497
25444 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25445 constants are representable in HOST_WIDE_INT.
25446 (parse_directive): Ditto.
25448 2017-04-25 Martin Sebor <msebor@redhat.com>
25451 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25452 (new_zero_array): Adjust signature.
25453 (dom_info::dom_init): Used unsigned rather that size_t.
25454 (dom_info::dom_info): Same.
25456 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25457 Jakub Jelinek <jakub@redhat.com>
25460 * config/arm/arm.c: Include gimple.h.
25461 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25462 returns negative, increment ncrn only if it returned positive.
25463 (arm_needs_doubleword_align): Return int instead of bool,
25464 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25465 members, but if there is any such non-FIELD_DECL
25466 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25467 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25468 returns negative, increment nregs only if it returned positive.
25469 (arm_setup_incoming_varargs): Likewise.
25470 (arm_function_arg_boundary): Emit -Wpsabi note if
25471 arm_needs_doubleword_align returns negative, return
25472 DOUBLEWORD_ALIGNMENT only if it returned positive.
25474 2017-04-25 Marek Polacek <polacek@redhat.com>
25477 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25478 first argument to type.
25480 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25483 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25484 type checks to test for compatibility instead of equality.
25486 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25487 Jakub Jelinek <jakub@redhat.com>
25490 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25492 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25493 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25494 the alignment computation, but return their maximum in warn_alignment.
25495 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25496 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25498 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25499 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25502 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25504 * config/arc/simdext.md (dmpyh): Fix typo.
25506 2017-04-25 Richard Biener <rguenther@suse.de>
25508 PR tree-optimization/80492
25509 * alias.c (compare_base_decls): Handle registers with asm
25510 specification conservatively.
25511 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25512 compare_base_decls returning dont-know properly.
25514 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25516 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25517 (legitimate_offset_address_p): New function.
25518 (arc_legitimate_address_p): Use above function.
25520 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25522 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25524 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25526 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25527 ACCH registers whenever they are available.
25529 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25531 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25532 double regs fix when not used.
25534 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25536 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25538 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25539 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25541 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25543 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25544 when emitting short ADD instructions.
25546 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25548 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25550 (cmpsi_cc_c_insn): Likewise.
25551 (cbranchsi4_scratch): Compute proper instruction length using
25552 compact_hreg_operand.
25553 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25555 2017-04-25 Richard Biener <rguenther@suse.de>
25557 PR middle-end/80509
25558 * passes.c (pass_manager::pass_manager): Initialize
25559 m_name_to_pass_map.
25561 2017-04-25 Richard Biener <rguenther@suse.de>
25563 PR tree-optimization/79201
25564 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25566 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25569 * config/s390/vector.md: Split MEM->GPR vector moves for
25570 non-s_operand addresses.
25572 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25575 * config/s390/predicates.md (reload_const_wide_int_operand): New
25577 * config/s390/s390.md ("movti"): Remove d/P alternative.
25578 ("movti_bigconst"): New pattern definition.
25580 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25583 * s390-protos.h (s390_expand_cs_hqi): Removed.
25584 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25585 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25586 modes as well as CCZ1mode and CCZmode.
25587 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25588 signature of s390_emit_compare_and_swap.
25589 (s390_expand_cs_hqi): Likewise, make static.
25590 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25591 compare-and-swap, in some cases.
25592 (s390_expand_cs): Wrapper function.
25593 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25595 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25596 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25597 patterns for small and large integers. Forbid symref memory operands.
25598 Move expander to s390.c. Require cc register.
25599 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25600 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25601 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25602 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25603 symref memory operands. Remove CC mode and call s390_match_ccmode
25605 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25607 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25609 * config/s390/s390.md (define_peephole2): New peephole to help
25610 combining the load-and-test pattern with volatile memory.
25612 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25614 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25615 with CCZmode for TARGET_Z196.
25617 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25619 PR rtl-optimization/80501
25620 * combine.c (make_compound_operation_int): Set subreg_code to SET
25621 even for AND with mask of the sign bit of mode.
25623 PR rtl-optimization/80500
25624 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25625 sum's initial value.
25627 2017-04-25 Julian Brown <julian@codesourcery.com>
25628 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25630 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25632 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25634 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25636 2017-04-25 Julian Brown <julian@codesourcery.com>
25637 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25639 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25640 (thunderx2t99_sha): New Reservation.
25642 2017-04-25 Julian Brown <julian@codesourcery.com>
25643 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25645 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25646 type for 1-element load.
25648 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25650 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25652 2017-04-24 Martin Jambor <mjambor@suse.cz>
25654 PR tree-optimization/80293
25655 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25656 char arrays not totally scalarizable if it is false.
25657 (analyze_all_variable_accesses): Pass correct value in the new
25658 parameter. Add a statistics counter.
25660 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25662 PR middle-end/79931
25663 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25665 2017-04-24 Richard Biener <rguenther@suse.de>
25667 PR tree-optimization/80494
25668 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25669 out for complex types.
25671 2017-04-24 Richard Biener <rguenther@suse.de>
25673 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25674 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25675 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25676 (DFS): Adjust and never fail.
25677 (sccvn_dom_walker::fail): Remove.
25678 (sccvn_dom_walker::before_dom_children): Adjust.
25679 (run_scc_vn): Likewise and never fail.
25680 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25681 (pass_fre::execute): Likewise.
25683 2017-04-24 Richard Biener <rguenther@suse.de>
25685 PR tree-optimization/79725
25686 * tree-ssa-sink.c (statement_sink_location): Return whether
25687 failure reason was zero uses. Move that check later.
25688 (sink_code_in_bb): Deal with zero uses by removing the stmt
25691 2017-04-24 Richard Biener <rguenther@suse.de>
25694 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25695 pointer-based references.
25697 2017-04-24 Richard Biener <rguenther@suse.de>
25700 * pass_manager.h (pass_manager::operator new): Remove.
25701 (pass_manager::operator delete): Likewise.
25702 * passes.c (pass_manager::operator new): Remove.
25703 (pass_manager::operator delete): Likewise.
25704 (pass_manager::pass_manager): Zero individual pass members.
25706 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25709 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25710 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25711 Check "XEXP (src, 1)" operand here.
25712 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25713 Check "XEXP (src, 1)" operand here.
25714 (dimode_scalar_chain::make_vector_copies): Detect count register
25715 of a shift instruction. Zero extend count register from QImode
25716 to DImode to satisfy vector shift pattern count operand predicate.
25717 Substitute vector shift count operand with a DImode copy.
25718 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25721 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25723 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25724 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25725 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25726 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25727 (UNSPEC_NOREX_MEM): Remove definition.
25729 2017-04-21 Richard Biener <rguenther@suse.de>
25731 PR tree-optimization/79547
25732 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25733 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25734 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25735 without any constraints.
25737 2017-04-21 Richard Biener <rguenther@suse.de>
25739 PR tree-optimization/78847
25740 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25742 2017-04-21 Richard Biener <rguenther@suse.de>
25744 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25745 (build_distinct_type_copy): Likewise.
25746 (build_variant_type_copy): Likewise.
25747 * tree.c (build_qualified_type): Pass down mem-stat info.
25748 (build_distinct_type_copy): Likewise.
25749 (build_variant_type_copy): Likewise.
25751 2017-04-21 Richard Biener <rguenther@suse.de>
25753 PR tree-optimization/80237
25754 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25756 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25757 for a simplified result.
25759 2016-04-21 Richard Biener <rguenther@suse.de>
25761 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25762 sth as strict as a simple_iv but a chrec without symbols and an
25763 operand defined in the loop we are peeling (and not some subloop).
25764 (propagate_constants_for_unrolling): Propagate all constants.
25766 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25769 * config/i386/i386.c (print_reg): Remove assert for disalowed
25770 regno values, call output_operand_lossage instead.
25772 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25775 * config/i386/constraints.md (Yc): New register constraint.
25776 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25777 Use Yc constraint for alternative 2 of operand 0. Remove
25778 preferred_for_speed attribute.
25780 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25782 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25783 lastprivate clauses in SIMT case.
25785 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25787 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25789 2017-04-20 Richard Biener <rguenther@suse.de>
25791 PR tree-optimization/57796
25792 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25793 as N scalar stores.
25794 (vect_model_load_cost): Cost gathers as N scalar loads.
25796 2017-04-20 Richard Biener <rguenther@suse.de>
25798 * ggc-page.c (ggc_allocated_p): Rename to ...
25799 (safe_lookup_page_table_entry): ... this and return the lookup
25801 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25803 2017-04-20 Richard Biener <rguenther@suse.de>
25805 PR tree-optimization/80453
25806 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25807 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25808 from the conditions.
25809 (vn_phi_eq): Pass them down.
25810 (vn_phi_lookup): Record them.
25811 (vn_phi_insert): Likewise.
25813 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25815 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25816 uninitialized variable warning to avoid buffer overrun.
25818 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25821 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25822 is suppressed for '{ 0 }' in C.
25824 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25826 * BASE-VER: Set to 8.0.0.
25828 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25830 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25831 priority .init_array and .fini_array section with SECTION_NOTYPE
25834 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25836 PR middle-end/80423
25837 * tree.h (build_array_type): Add typeless_storage default argument.
25838 * tree.c (type_cache_hasher::equal): Also compare
25839 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25840 (build_array_type): Add typeless_storage argument, set
25841 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25843 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25844 (build_array_type): Likewise. Add typeless_storage argument.
25846 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25847 Jakub Jelinek <jakub@redhat.com>
25849 PR tree-optimization/80426
25850 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25851 operation on symbolic operands, also compute the overflow for the
25852 invariant part when the operation degenerates into a negation.
25854 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25857 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25858 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25861 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25863 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25866 * config/avr/avr.c (tree.h): Include it.
25867 (cgraph.h): Include it.
25868 (avr_encode_section_info): Don't warn for uninitialized progmem
25869 variable if it's just an alias.
25871 2017-04-19 Richard Biener <rguenther@suse.de>
25874 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25875 when needed by AutoPGO.
25877 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25880 * doc/lto.texi: Remove an extra 'that'.
25882 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25884 PR rtl-optimization/80429
25885 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25886 are only used in debug insns.
25888 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25889 Vladimir Makarov <vmakarov@redhat.com>
25891 * config/sparc/predicates.md (input_operand): Add comment. Return
25892 true for any memory operand when LRA is in progress.
25893 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25895 2017-04-18 Jeff Law <law@redhat.com>
25898 * mips.md ({return,simple_return}_internal): Do not overwrite
25901 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25903 PR tree-optimization/80443
25904 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25905 instead of adding 1, subtract -1 and similarly instead of subtracting
25908 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25910 PR rtl-optimization/80357
25911 * haifa-sched.c (tmp_bitmap): New variable.
25912 (model_recompute): Handle duplicate use records.
25913 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25914 (free_global_sched_pressure_data): Free it.
25916 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25919 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25920 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25921 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25922 instead of SYSTEM_HEADER_DIR.
25924 2017-04-18 Jeff Law <law@redhat.com>
25926 PR middle-end/80422
25927 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25928 predecessors after walking up the insn chain.
25930 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25933 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25934 sizetype type into debug info.
25936 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25939 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25940 unneeded test for TARGET_UPPER_REGS_SF.
25941 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25943 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25946 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25947 instead of gsi_after_labels.
25949 2017-04-18 Jeff Law <law@redhat.com>
25951 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25955 2017-04-13 Jeff Law <law@redhat.com>
25956 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25957 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25959 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25962 * config/avr/avr.c (intl.h): Include it.
25963 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25965 2017-04-18 Martin Liska <mliska@suse.cz>
25967 PR gcov-profile/78783
25968 * gcov-tool.c (gcov_output_files): Validate that destination
25969 file is either removed by the tool or by a user.
25971 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25972 Guy Benyei <guybe@mellanox.com>
25974 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25975 block, and do not negate it, the stored id is already negative.
25977 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25979 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25981 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25984 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25985 masks of options that should be turned off if the VSX vector
25986 options are turned off.
25987 (OTHER_P8_VECTOR_MASKS): Likewise.
25988 (OTHER_VSX_VECTOR_MASKS): Likewise.
25989 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25990 rs6000_disable_incompatible_switches to validate no type switches
25992 (rs6000_incompatible_switch): New function to disallow turning on
25993 other vector options if -mno-vsx, -mno-power8-vector, or
25994 -mno-power9-vector are specified.
25996 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25998 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26000 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26002 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26003 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26004 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26005 (ARG_POINTER_CFA_OFFSET): Likewise.
26007 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26009 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26010 conditions to take advantage of various optimizations.
26012 2017-04-13 Jeff Law <law@redhat.com>
26014 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26015 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26016 (zero_extendsidi2_dext): Likewise.
26018 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26021 * fold-const.c (fold_ternary_loc): Revert
26022 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26025 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26027 PR rtl-optimization/80343
26028 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26029 new scratch pseudo.
26031 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26034 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26035 to ubsan_encode_value.
26037 2017-04-13 Jeff Law <law@redhat.com>
26039 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26040 appearing in DEBUG_INSNs.
26042 2017-04-13 Martin Liska <mliska@suse.cz>
26044 PR gcov-profile/80413
26045 * gcov-io.c (gcov_write_string): Copy to buffer just when
26046 allocated size is greater than zero.
26048 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26051 * dwarf2out.c (decls_for_scope): Ignore declarations of
26052 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26054 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26057 * ipa-visibility.c (non_local_p): Fix typos.
26058 (localize_node): When localizing symbol in same comdat group,
26059 dissolve the group only when we know external symbols are going
26061 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26063 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26065 PR tree-optimization/79390
26066 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26067 order does not result in usable sequence, retry with reversed operand
26073 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26074 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26075 op0 instead of fold_convert_loc (loc, type, arg0).
26077 2017-04-12 Jeff Law <law@redhat.com>
26079 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26080 has a delay slot in the generated code.
26082 * config/cris/cris.md (cris_preferred_reload_class): Return
26083 GENNONACR_REGS rather than GENERAL_REGS.
26085 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26088 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26089 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26090 signedness of the result type.
26092 2017-04-12 Richard Biener <rguenther@suse.de>
26093 Jeff Law <law@redhat.com>
26095 PR tree-optimization/80359
26096 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26097 trim stores to TARGET_MEM_REFs.
26099 2017-04-12 Richard Biener <rguenther@suse.de>
26101 PR tree-optimization/79390
26102 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26103 threading case even more.
26105 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26108 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26109 for quad_address_p for TImode, instead of just not indexed_address.
26111 2017-04-12 Richard Biener <rguenther@suse.de>
26112 Bernd Edlinger <bernd.edlinger@hotmail.de>
26114 PR middle-end/79671
26115 * alias.c (component_uses_parent_alias_set_from): Handle
26116 TYPE_TYPELESS_STORAGE.
26117 (get_alias_set): Likewise.
26118 * tree-core.h (tree_type_common): Add typeless_storage flag.
26119 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26120 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26121 for types containing members with TYPE_TYPELESS_STORAGE.
26122 (place_field): Likewise.
26123 (layout_type): Likewise for ARRAY_TYPE.
26124 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26125 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26126 TYPE_TYPELESS_STORAGE.
26127 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26129 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26132 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26133 first argument to type.
26135 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26139 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26140 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26141 (rs6000_expand_binop_builtin): Likewise.
26142 (rs6000_expand_ternop_builtin): Likewise; also add missing
26143 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26144 vshasigma built-ins.
26145 * doc/extend.texi: Document that vec_xxpermdi's third argument
26146 must be a constant.
26148 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26150 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26151 Use shift_const cost parameter when calculating gain of STV shifts.
26153 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26155 PR rtl-optimization/70478
26156 * lra-constraints.c (process_alt_operands): Check memory for
26157 disfavoring memory insn operand.
26159 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26161 PR middle-end/80100
26162 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26163 left shift in unsigned HOST_WIDE_INT type.
26165 PR rtl-optimization/80385
26166 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26167 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26170 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26171 if they have any depend clauses.
26173 2017-04-11 Martin Liska <mliska@suse.cz>
26176 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26177 * ipa-split.c (split_function): Create a local comdat symbol
26178 if caller is in a comdat group.
26180 2017-04-11 Martin Liska <mliska@suse.cz>
26183 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26186 2017-04-11 Martin Sebor <msebor@redhat.com>
26188 PR middle-end/80364
26189 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26190 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26192 (directive::set_width, directive::set_precision, format_character):
26194 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26197 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26200 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26201 conflict, set target->arch_name instead of target->cpu_name.
26203 2017-04-11 Richard Biener <rguenther@suse.de>
26205 PR tree-optimization/80374
26206 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26207 build_zero_cst, remove fold_convertible_p check again.
26209 2017-04-11 Martin Liska <mliska@suse.cz>
26212 * ubsan.c (instrument_object_size): Do not instrument register
26215 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26218 * config/i386/i386-builtin-types.def
26219 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26220 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26221 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26222 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26223 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26224 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26225 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26226 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26227 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26228 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26229 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26230 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26231 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26232 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26233 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26234 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26235 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26236 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26237 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26238 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26239 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26240 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26241 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26242 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26243 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26244 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26245 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26246 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26247 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26248 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26249 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26250 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26251 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26252 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26253 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26254 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26255 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26256 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26257 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26258 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26259 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26260 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26261 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26262 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26263 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26264 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26266 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26267 flag to second_arg_count, handle 4 argument function type _COUNT
26268 aliases, handle second_arg_count on second argument rather than last.
26270 2017-04-10 Jeff Law <law@redhat.com>
26272 PR tree-optimization/80374
26273 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26274 record anything if we can not convert integer_zero_node to the
26277 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26280 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26281 Enhance special handling given to the TARGET_P9_MINMAX option in
26282 relation to certain other options.
26284 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26286 PR tree-optimization/80153
26287 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26288 remove POINTER_PLUS_EXPR's base part directly, rather than through
26291 2017-04-10 Richard Biener <rguenther@suse.de>
26292 Bin Cheng <bin.cheng@arm.com>
26294 PR tree-optimization/80153
26295 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26296 the first element of pointer type aff_tree. Build result expr in
26298 (add_elt_to_tree): Convert to type unconditionally. Remove other
26299 fold_convert calls.
26300 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26301 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26303 2017-04-10 Richard Biener <rguenther@suse.de>
26305 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26308 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26310 PR rtl-optimization/70478
26311 * lra-constraints.c (curr_small_class_check): New.
26312 (update_and_check_small_class_inputs): New.
26313 (process_alt_operands): Update curr_small_class_check. Disfavor
26314 alternative insn memory operands. Check available regs for small
26317 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26320 * config/mips/mips.opt (-mvirt): Update description.
26321 * doc/invoke.texi (-mvirt): Likewise.
26323 2017-04-10 Richard Biener <rguenther@suse.de>
26325 PR middle-end/80362
26326 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26327 looking for NEGATE_EXPR in -A / -B to A / B folding.
26329 2017-04-10 Martin Liska <mliska@suse.cz>
26331 PR gcov-profile/80224
26332 * gcov.c (print_usage): Fix usage string.
26333 (get_gcov_intermediate_filename): Remove.
26334 (output_gcov_file): Use both for normal and intermediate format.
26335 (generate_results): Do not initialize special file for
26336 intermediate format.
26338 2017-04-10 Richard Biener <rguenther@suse.de>
26340 PR tree-optimization/80304
26341 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26344 2017-04-10 Nathan Sidwell <nathan@acm.org>
26347 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26348 (rs6000_init_builtins): Use it.
26350 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26352 * config/arm/arm.md (<mrc>): Add mode to SET source.
26353 (<mrrc>): Likewise.
26355 2017-04-10 Richard Biener <rguenther@suse.de>
26357 PR middle-end/80344
26358 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26360 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26363 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26364 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26365 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26366 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26367 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26368 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26369 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26370 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26371 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26372 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26373 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26374 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26375 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26376 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26377 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26378 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26379 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26380 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26381 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26382 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26383 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26384 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26385 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26387 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26389 PR rtl-optimization/70478
26390 * lra-constraints.c: Reverse the last patch.
26392 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26394 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26395 Add comment for WCHAR_T.
26397 2017-04-08 Martin Liska <mliska@suse.cz>
26400 2017-04-07 Martin Liska <mliska@suse.cz>
26403 * ipa-split.c (split_function): Add function part to a same comdat
26406 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26409 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26411 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26413 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26414 (vec_pairing): ...to this.
26415 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26416 (rs6000_sched_init): Adjust for name change.
26417 (struct rs6000_sched_context): Likewise.
26418 (rs6000_init_sched_context): Likewise.
26419 (rs6000_set_sched_context): Likewise.
26421 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26427 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26429 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26430 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26431 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26433 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26435 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26437 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26439 PR rtl-optimization/70703
26440 * ira-color.c (update_conflict_hard_regno_costs): Use
26441 int64_t instead of HOST_WIDE_INT.
26443 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26445 PR rtl-optimization/70478
26446 * lra-constraints.c (process_alt_operands): Disfavor alternative
26447 insn memory operands.
26449 2017-04-07 Jeff Law <law@redhat.com>
26451 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26452 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26454 2017-04-07 Martin Liska <mliska@suse.cz>
26457 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26458 Show error message instead of an ICE.
26460 2017-04-07 Martin Liska <mliska@suse.cz>
26463 * ipa-split.c (split_function): Add function part to a same comdat
26466 2017-04-07 Richard Biener <rguenther@suse.de>
26468 PR middle-end/80341
26469 * tree.c (get_unwidened): Also handle ! for_type case for
26471 * convert.c (do_narrow): Split out from ...
26472 (convert_to_integer_1): ... here. Do not pass final truncation
26473 type to get_unwidened for TRUNC_DIV_EXPR.
26475 2017-04-07 Richard Biener <rguenther@suse.de>
26477 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26479 (aff_combination_const): Adjust.
26480 (aff_combination_scale): Likewise.
26481 (aff_combination_add_elt): Likewise.
26482 (aff_combination_add_cst): Likewise.
26483 (aff_combination_convert): Likewise.
26484 (add_elt_to_tree): Likewise. Remove unused argument.
26485 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26487 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26489 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26491 * config/arm/arm.c (arm_default_short_enums): Use
26492 ARM_DEFAULT_SHORT_ENUMS.
26493 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26495 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26498 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26499 members with redundant out-of-class redeclaration.
26501 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26504 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26505 * config/i386/i386.md (*zero_extendsidi2):
26506 Add (?*x,*x) and (?*v,*v) alternatives.
26508 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26511 * config/i386/i386.c (ix86_expand_builtin)
26512 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26513 mode from insn data. Convert operands to insn operand mode.
26514 Copy operands that don't satisfy insn predicate to a register.
26516 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26518 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26521 2017-04-06 Richard Biener <rguenther@suse.de>
26523 PR tree-optimization/80334
26524 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26525 preserve alignment of accesses.
26527 2017-04-06 Richard Biener <rguenther@suse.de>
26529 PR tree-optimization/80262
26530 * tree-sra.c (build_ref_for_offset): Preserve address-space
26532 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26533 Drop useless address-space information on MEM_REF offsets.
26535 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26537 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26539 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26541 PR rtl-optimization/70703
26542 * ira-color.c (update_conflict_hard_regno_costs): Use
26543 HOST_WIDE_INT instead of long.
26545 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26548 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26549 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26551 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26552 for x86_64 target. Handle -m3dnowa option.
26554 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26556 PR rtl-optimization/70703
26557 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26558 (update_conflict_hard_regno_costs): Use long instead of unsigned
26559 arithmetic for cost calculation.
26561 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26562 Bernd Edlinger <bernd.edlinger@hotmail.de>
26565 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26568 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26571 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26572 ptr_mode with Pmode throughout.
26573 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26574 into probe_stack_range and use DImode.
26576 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26579 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26580 call_eh_return is true.
26582 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26584 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26585 Initialize last_match_fntype_index.
26587 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26590 * tree-nvr.c: Include internal-fn.h.
26591 (pass_return_slot::execute): Ignore internal calls without
26594 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26595 Richard Biener <rguenther@suse.de>
26598 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26599 captures used multiple times, except for the last use.
26600 * generic-match-head.c: Include gimplify.h.
26602 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26604 PR tree-optimization/79390
26605 * target.h (struct noce_if_info): Declare.
26606 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26607 * target.def (noce_conversion_profitable_p): New target hook.
26608 * ifcvt.h (struct noce_if_info): New type, moved from ...
26609 * ifcvt.c (struct noce_if_info): ... here.
26610 (noce_conversion_profitable_p): Renamed to ...
26611 (default_noce_conversion_profitable_p): ... this. No longer
26613 (noce_try_store_flag_constants, noce_try_addcc,
26614 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26615 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26616 instead of noce_conversion_profitable_p.
26617 * config/i386/i386.c: Include ifcvt.h.
26618 (ix86_option_override_internal): Don't override
26619 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26620 (ix86_noce_conversion_profitable_p): New function.
26621 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26622 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26623 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26624 * doc/tm.texi: Regenerated.
26626 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26628 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26631 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26634 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26635 instructions for small multiply cores.
26637 2017-04-04 Jeff Law <law@redhat.com>
26639 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26641 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26642 that are not set by the loop over the elements.
26644 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26647 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26648 int mode, convert_modes it to mode as unsigned, otherwise use
26649 lowpart_subreg to mode rather than SImode.
26650 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26651 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26652 Use DImode instead of SImode for the shift count operand.
26653 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26656 2017-04-04 Richard Biener <rguenther@suse.de>
26658 PR middle-end/80281
26659 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26660 arithmetic done for the negate or the plus. Simplify.
26661 (A - (-B) -> A + B): Likewise.
26662 * fold-const.c (split_tree): Make sure to not negate pointers.
26664 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26666 PR rtl-optimization/60818
26667 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26668 a compare of comparisons with the thing compared if this results
26669 in a different machine mode.
26671 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26673 * alias.c (base_alias_check): Fix typo in comment.
26674 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26675 * cgraphunit.c (symbol_table::compile): Likewise.
26676 * collect2.c (maybe_run_lto_and_relink): Likewise.
26677 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26678 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26679 * config/avr/avr.c (avr_map_op_t): Likewise.
26680 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26681 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26682 * config/epiphany/epiphany.md (movcc): Likewise.
26683 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26684 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26686 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26687 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26688 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26689 * config/sh/sh.c (sh_rtx_costs): Likewise.
26690 * fold-const.c (fold_truth_andor): Likewise.
26691 * genautomata.c (collapse_flag): Likewise.
26692 * gengtype.h (struct type::u::s): Likewise.
26693 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26694 * input.c (FORMAT_AMOUNT): Likewise.
26695 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26696 (known_aggs_to_agg_replacement_list): Likewise.
26697 * ipa-inline-analysis.c: Likewise.
26698 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26699 * ipa-polymorphic-call.c
26700 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26701 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26702 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26704 * modulo-sched.c (apply_reg_moves): Likewise.
26705 * omp-expand.c (build_omp_regions_1): Likewise.
26706 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26707 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26708 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26709 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26710 * value-prof.c: Likewise.
26711 * var-tracking.c (val_reset): Likewise.
26713 2017-04-03 Richard Biener <rguenther@suse.de>
26715 PR tree-optimization/80275
26716 * fold-const.c (split_address_to_core_and_offset): Handle
26719 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26721 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26722 descriptors is at least equal to that of functions.
26724 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26726 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26728 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26731 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26732 (mov<IMOD4:mode>): New expander.
26733 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26735 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26737 PR rtl-optimization/79405
26738 * fwprop.c (propagations_left): New variable.
26739 (forward_propagate_into): Decrement it.
26740 (fwprop_init): Initialize it.
26741 (fw_prop): If the variable has reached zero, stop propagating.
26742 (fwprop_addr): Ditto.
26744 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26747 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26748 a FUNCTION_DECL, pass it as decl instead of origin to
26751 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26753 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26756 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26759 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26760 TARGET_VSX_SMALL_INTEGER.
26762 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26764 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26765 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26767 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26769 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26770 extraction from odd-numbered MSA register.
26772 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26774 PR middle-end/80173
26775 * expmed.c (store_bit_field_1): Don't attempt to create
26776 a word subreg out of hard registers wider than word if they
26777 have HARD_REGNO_NREGS of 1 for their mode.
26779 PR middle-end/80163
26780 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26781 conversions to integer types wider than word and pointer.
26784 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26785 (rtx_equal_for_cselib_p): Pass 0 to it.
26786 * cselib.c (cselib_hasher::equal): Likewise.
26787 (rtx_equal_for_cselib_1): Add depth argument. If depth
26788 is 128, don't look up VALUE locs and punt. Increment
26789 depth in recursive calls when walking VALUE locs.
26791 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26793 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26794 (make_gcov_file_name): Use the canonical path name for generating
26796 (read_line): Fix handling of files with ascii null bytes.
26798 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26800 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26801 to initialise a vector register instead
26802 of using a const_int.
26804 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26806 PR translation/80189
26807 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26808 diagnostic messages.
26810 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26813 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26814 (dfp_diex_<mode>): Update mode of operand 1.
26815 * doc/extend.texi (dxex, dxexq): Document change to return type.
26816 (diex, diexq): Document change to argument type.
26818 2017-03-30 Martin Jambor <mjambor@suse.cz>
26821 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26822 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26823 it reflects the signature changes performed at the callee side.
26824 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26825 to cgraph_build_function_type_skip_args.
26826 (build_function_decl_skip_args): Adjust call to the above function.
26828 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26831 * config/i386/sse.md
26832 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26833 register as dest whenever it is a MEM not rtx_equal_p to the
26834 corresponding dup operand, and when forcing into reg move the
26835 reg into the memory afterwards.
26836 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26837 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26838 for the force_reg mode.
26839 (avx512vl_vextractf128<mode>): Use register as dest either
26840 always when a MEM, or when it is a MEM not rtx_equal_p to the
26841 corresponding dup operand, or even not when it is a CONST_VECTOR
26842 depending on the mode and lo vs. hi.
26843 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26845 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26846 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26847 Likewise. Require that operands[2] is even.
26848 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26849 Remove extraneous parens. Require that operands[2] is a multiple
26851 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26852 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26853 disallow memory then.
26855 2017-03-30 Richard Biener <rguenther@suse.de>
26857 PR tree-optimization/77498
26858 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26859 to non-constants over backedges.
26861 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26863 PR rtl-optimization/80233
26864 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26865 as last_combined_insn. Do not test for BARRIER_P separately.
26867 2017-03-29 Andreas Schwab <schwab@suse.de>
26870 * calls.c (prepare_call_address): Convert funexp to Pmode before
26871 copying to temp reg.
26873 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26875 PR tree-optimization/80158
26876 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26877 Handle possible future case of more than one alternate
26879 (replace_rhs_if_not_dup): Likewise.
26880 (replace_one_candidate): Likewise.
26882 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26884 PR rtl-optimization/80193
26885 * ira.c (ira): Do not check allocation for LRA.
26887 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26889 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26890 (nvptx_output_simt_exit): Declare.
26891 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26892 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26893 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26894 (nvptx_declare_function_name): ...here. Emit declaration of local
26895 memory space buffer for omp_simt_enter insn.
26896 (nvptx_output_unisimt_switch): New.
26897 (nvptx_output_softstack_switch): New.
26898 (nvptx_output_simt_enter): New.
26899 (nvptx_output_simt_exit): New.
26900 * config/nvptx/nvptx.h (struct machine_function): New fields
26901 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26902 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26903 (UNSPECV_SIMT_EXIT): Ditto.
26904 (omp_simt_enter_insn): New insn.
26905 (omp_simt_enter): New expansion.
26906 (omp_simt_exit): New insn.
26907 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26909 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26910 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26911 (expand_GOMP_SIMT_EXIT): New.
26912 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26913 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26914 (GOMP_SIMT_EXIT): Ditto.
26915 * target-insns.def (omp_simt_enter): New insn.
26916 (omp_simt_exit): Ditto.
26917 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26919 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26920 (lower_rec_input_clauses): Likewise.
26921 (lower_lastprivate_clauses): Handle SIMT privatization.
26923 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26924 (ompdevlow_adjust_simt_enter): New.
26925 (find_simtpriv_var_op): New.
26926 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26927 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26929 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26930 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26931 (copy_decl_for_dup_finish): Ditto.
26933 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26935 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26938 * config/i386/i386.c (ix86_option_override_internal): Always
26939 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26941 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26943 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26945 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26947 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26948 mark new edge's irreducible flag accordign to it.
26949 (vect_do_peeling): Check loop preheader edge's irreducible flag
26950 and pass it to function slpeel_add_loop_guard.
26952 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26954 PR tree-optimization/80218
26955 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26956 Update block frequencies and counts.
26958 2017-03-28 Richard Biener <rguenther@suse.de>
26960 PR tree-optimization/78644
26961 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26962 of a simplification result we may not use it at all.
26964 2017-03-28 Richard Biener <rguenther@suse.de>
26967 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26968 without arguments, generate default definition of a SSA name.
26970 2017-03-28 Richard Biener <rguenther@suse.de>
26972 PR middle-end/80222
26973 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26974 TYPE_REF_CAN_ALIAS_ALL references.
26975 * fold-const.c (fold_indirect_ref_1): Likewise.
26977 2017-03-28 Martin Liska <mliska@suse.cz>
26980 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26981 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26983 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26984 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26986 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26987 (EXTRA_SPECS): Define.
26988 (SUBTARGET_EXTRA_SPECS): Likewise.
26989 (SUBTARGET_CPP_SPEC): Likewise.
26990 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26991 SUBTARGET_EXTRA_SPECS.
26992 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26994 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26996 * config/arc/simdext.md (vst64_insn): Update pattern.
26997 (vld32wh_insn): Likewise.
26998 (vld32wl_insn): Likewise.
26999 (vld64_insn): Likewise.
27000 (vld32_insn): Likewise.
27002 2017-03-28 Marek Polacek <polacek@redhat.com>
27005 * fold-const.c (fold_comparison): Use protected_set_expr_location
27006 instead of SET_EXPR_LOCATION.
27008 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27010 * tree.c (add_expr): Avoid name lookup warning.
27012 2017-03-27 Jeff Law <law@redhat.com>
27014 PR tree-optimization/80216
27015 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27016 function name. Limit recursion depth.
27017 (record_temporary_equivalences): Corresponding changes.
27019 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27021 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27024 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27027 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27029 * cfgcleanup.c (reg_note_cfa_p): New array.
27030 (insns_have_identical_cfa_notes): New function.
27031 (old_insns_match_p): Don't cross-jump in between /f
27032 and non-/f instructions. If both i1 and i2 are frame related,
27033 verify all CFA notes, their order and content.
27035 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27038 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27039 HImode and SImode with zero extend to DImode to one insn.
27040 (bswap<mode>2_extenddi): Likewise.
27041 (bswapsi2_extenddi): Likewise.
27042 (bswaphi2_extendsi): Likewise.
27043 (bswaphi2): Combine bswap HImode and SImode into one insn.
27044 Separate memory insns from swapping register.
27045 (bswapsi2): Likewise.
27046 (bswap<mode>2): Likewise.
27047 (bswaphi2_internal): Delete, no longer used.
27048 (bswapsi2_internal): Likewise.
27049 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27050 store, and gpr<-gpr swap insns.
27051 (bswap<mode>2_store): Likewise.
27052 (bswaphi2_reg): Register only splitter, combine with the splitter.
27053 (bswaphi2 splitter): Likewise.
27054 (bswapsi2_reg): Likewise.
27055 (bswapsi2 splitter): Likewise.
27056 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27057 the insns into load, store, and register/register insns.
27058 (bswapdi2_ldbrx): Likewise.
27059 (bswapdi2_load): Likewise.
27060 (bswapdi2_store): Likewise.
27061 (bswapdi2_reg): Likewise.
27063 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27065 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27066 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27068 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27071 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27073 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27074 special handling for target option conflicts between dform
27075 options (-mpower9-dform, -mpower9-dform-vector,
27076 -mpower9-dform-scalar) and -mno-direct-move.
27078 2017-03-27 Richard Biener <rguenther@suse.de>
27080 PR tree-optimization/80181
27081 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27083 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27085 * config/arc/predicates.md (move_double_src_operand): Replace the
27086 call to move_double_src_operand with a call to address_operand.
27088 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27090 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27091 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27092 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27094 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27096 * config/arc/predicates.md (long_immediate_loadstore_operand):
27097 Consider scaled addresses cases.
27099 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27101 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27102 restored when in interrupt.
27103 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27104 doesn't have delay slot.
27106 2017-03-27 Richard Biener <rguenther@suse.de>
27109 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27110 inlined thunk clones.
27112 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27115 * asan.c (instrument_derefs): Copy over last operand from
27116 original COMPONENT_REF to the new COMPONENT_REF with
27117 DECL_BIT_FIELD_REPRESENTATIVE.
27118 * ubsan.c (instrument_object_size): Likewise.
27120 2017-03-27 Richard Biener <rguenther@suse.de>
27122 PR tree-optimization/80170
27123 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27124 sure DR/SCEV didnt fold in constants we do not see when looking
27125 at the reference base alignment.
27127 2017-03-27 Richard Biener <rguenther@suse.de>
27129 PR middle-end/80171
27130 * gimple-fold.c (fold_ctor_reference): Properly guard against
27131 NULL return value from canonicalize_constructor_val.
27133 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27136 * config/i386/i386.c (ix86_expand_builtin)
27137 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27138 flags reg setting and flags reg using instructions.
27139 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27140 clobbering instructions to zero extend op2.
27142 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27144 * doc/install.texi (Configuration) <--with-aix-soname>:
27145 Update link to AIX ld.
27147 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27149 PR rtl-optimization/80160
27150 PR rtl-optimization/80159
27151 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27152 reg_alternate_class into account.
27154 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27157 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27158 to consider in curr_insn_transform.
27160 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27162 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27163 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27164 and emit_mode_inner.
27166 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27168 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27169 argument to the overloaded builtin variants. Use the new flag to
27170 deprecate certain builtin variants.
27171 * config/s390/s390-builtin-types.def: Add new builtin types.
27172 * config/s390/s390-builtins.h: Support new flags field for
27173 overloaded builtins.
27174 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27175 (s390_macro_to_expand): Enable vector float data type.
27176 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27177 builtins by incrementing the __VEC__ version number.
27178 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27180 (s390_resolve_overloaded_builtin): Emit error messages depending
27181 on the builtin flags.
27182 * config/s390/s390.c (s390_expand_builtin): Support additional
27183 flags argument. Change error message to match the messages
27184 emitted in s390-c.c.
27185 * config/s390/s390.md: New UNSPEC_* constants.
27186 (op_type): Add new instruction types.
27187 * config/s390/vecintrin.h: Add new builtins and test data class
27189 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27190 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27191 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27192 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27193 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27194 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27196 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27197 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27198 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27199 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27201 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27202 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27203 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27204 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27206 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27207 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27208 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27209 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27210 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27211 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27212 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27214 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27215 ("vec_scatter_element<V_HW_4:mode>_DI")
27216 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27217 ("vec_fpint<mode>", "vflls")
27218 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27219 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27220 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27221 ("*vec_cmphe<mode>_cc"): ... these.
27223 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27224 mode constant instead of magic value.
27226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27228 * config/s390/s390.c (s390_expand_vec_compare): Support other
27229 vector floating point modes than just V2DF.
27230 (s390_expand_vcond): Likewise.
27231 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27232 (s390_cannot_change_mode_class): Prevent mode changes between TF
27233 and V1TF in vector registers.
27234 * config/s390/s390.md (DF, SF): New mode attributes.
27235 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27236 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27237 SFmode support for VRs.
27238 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27240 (VFT, VF_HW): New mode iterators.
27241 (vw, sdx): New mode attributes.
27242 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27243 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27244 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27245 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27246 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27247 also the new vector floating point modes. Renaming to ...
27249 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27250 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27251 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27252 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27253 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27254 ("vec_unordered<mode>"): ... these.
27256 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27257 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27258 ("*vec_extendv2df"): New insn definitions.
27260 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27262 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27263 ("mulditi3_2", "*muldi3_sign"): New patterns.
27264 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27265 rename the pattern definition.
27267 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27269 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27271 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27273 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27275 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27276 instruction if possible.
27277 * config/s390/vector.md (vec_halfnumelts): New mode
27279 ("*vec_vllezlf<mode>"): New pattern.
27281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27283 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27284 ("popcountv4si2", "popcountv2di2"): Rename to ...
27285 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27286 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27288 ("popcount<mode>2_vxe"): New pattern.
27290 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27292 * common/config/s390/s390-common.c (processor_flags_table): Add
27294 * config.gcc: Add arch12.
27295 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27296 Default to arch12 for unknown CPU model numbers.
27297 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27298 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27299 PROCESSOR_max sanity check.
27300 * config/s390/s390-opts.h (enum processor_type): Add
27302 * config/s390/s390.c (processor_table): Add arch12.
27303 (s390_expand_builtin): Add check for B_VXE flag.
27304 (s390_issue_rate): Add PROCESSOR_ARCH12.
27305 (s390_get_sched_attrmask): Likewise.
27306 (s390_get_unit_mask): Likewise.
27307 (s390_sched_score): Enable z13 scheduling for arch12.
27308 (s390_sched_reorder): Likewise.
27309 (s390_sched_variable_issue): Likewise.
27310 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27312 (s390_tune_attr): Use z13 scheduling also for arch12.
27313 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27314 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27315 (TARGET_VXE_P): New macros.
27316 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27317 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27318 * config/s390/s390.opt: Add arch12 as processor_type.
27320 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27322 * config/s390/s390.md
27323 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27324 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27325 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27327 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27328 Rename expanders to ...
27330 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27331 ("fixuns_truncdddi2_emu"): ... these.
27333 ("fixuns_trunc<mode>si2_emu"): New expander.
27335 ("*fixuns_truncdfdi2_z13"): Rename to ...
27336 ("*fixuns_truncdfdi2_vx"): ... this.
27338 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27340 * config/s390/2964.md: Remove the single element vector compare
27341 instructions which are no longer used.
27342 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27344 (s390_canonicalize_comparison): Remove handling of DFmode
27346 (s390_expand_vec_compare_scalar): Remove function.
27347 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27348 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27350 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27352 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27354 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27356 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27357 will anyway by matched by mov<mode>_64dfp.
27359 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27361 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27362 vlef/vstef. Add missing operand to vleif.
27364 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27366 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27367 pair for all vector types with 64 bit elements.
27368 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27369 * config/s390/vector.md (V_HW_64): ... here.
27370 (V_128_NOSINGLE): New mode iterator.
27371 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27372 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27373 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27374 ("*vec_load_pairv2di"): Change to ...
27375 ("*vec_load_pair<mode>"): ... this one.
27377 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27379 * config/s390/constraints.md: Add comments.
27380 (jKK): Reject element sizes > 8 bytes.
27381 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27383 * config/s390/s390.md: Add the s_operand checks formerly in
27384 s390_split_ok_p to various splitters where they are still
27386 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27387 for 128 bit vectors. Plus two splitters.
27389 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27391 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27394 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27397 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27398 error if the boundary argument is not constant.
27400 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27402 PR rtl-optimization/80112
27403 * loop-doloop.c (doloop_condition_get): Don't check condition
27404 if cmp isn't SET with IF_THEN_ELSE src.
27406 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27408 PR tree-optimization/80158
27409 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27410 replacing a candidate statement, also replace it for the
27411 candidate's alternate interpretation.
27412 (replace_rhs_if_not_dup): Likewise.
27413 (replace_one_candidate): Likewise.
27415 2017-03-24 Richard Biener <rguenther@suse.de>
27417 PR tree-optimization/80167
27418 * graphite-isl-ast-to-gimple.c
27419 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27421 (translate_isl_ast_to_gimple::get_rename): Likewise.
27423 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27425 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27426 handling of certain combinations of target options, including the
27427 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27428 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27430 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27433 * config/arm/arm.md (*load_multiple): Add reload_completed to
27434 matching condition.
27436 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27437 Richard Biener <rguenth@suse.de>
27439 PR tree-optimization/79908
27440 PR tree-optimization/80136
27441 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27442 been cast away, gimplify_and_add suffices.
27444 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27446 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27448 2017-03-23 Richard Biener <rguenther@suse.de>
27450 PR tree-optimization/80032
27451 * gimplify.c (gimple_push_cleanup): Forced unconditional
27452 cleanups still have to go to the conditional_cleanups
27455 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27457 PR tree-optimization/80072
27458 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27460 (next_operand_entry_id): Change type to unsigned int.
27461 (sort_by_operand_rank): Make sure to return the right return value
27462 even if unsigned fields are bigger than INT_MAX.
27463 (struct oecount): Change cnt and id type to unsigned int.
27464 (oecount_hasher::equal): Formatting fix.
27465 (oecount_cmp): Make sure to return the right return value
27466 even if unsigned fields are bigger than INT_MAX.
27467 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27470 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27471 TREE_READONLY on result if writing it more than once.
27474 * doc/invoke.texi (-fsanitize=thread): Document that with
27475 -fnon-call-exceptions atomics are not able to throw
27479 * tsan.c: Include tree-eh.h.
27480 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27481 maybe_clean_or_replace_eh_stmt where needed.
27482 (instrument_memory_accesses): Add cfg_changed argument.
27483 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27484 if it returned true.
27485 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27487 PR rtl-optimization/63191
27488 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27489 wrapper function, moved the whole old content into ...
27490 (ix86_delegitimize_address_1): ... this. New inline function.
27491 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27492 true as last argument instead of ix86_delegitimize_address.
27494 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27496 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27497 cortexa57_branch_cost.
27499 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27501 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27503 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27506 * doc/md.texi (Constraints): Document wA constraint.
27507 * config/rs6000/constraints.md (wA): New.
27508 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27509 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27510 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27511 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27513 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27516 * gimplify.c (is_oacc_declared): New function.
27517 (oacc_default_clause): Use it to set default flags for acc declared
27518 variables inside parallel regions.
27519 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27520 declared variables.
27521 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27522 declare attribute to any decl as necessary.
27524 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27527 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27528 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27529 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27530 (arm_arch_lpae): This.
27531 * config/arm/arm.c (arm_arch7ve): Rename into ...
27532 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27533 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27536 2017-03-22 Martin Liska <mliska@suse.cz>
27539 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27540 error message instead of an ICE.
27542 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27544 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27546 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27548 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27550 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27553 2017-03-21 Martin Sebor <msebor@redhat.com>
27555 * doc/extend.texi: Use "cannot" instead of "can't."
27556 * doc/hostconfig.texi: Same.
27557 * doc/install.texi: Same.
27558 * doc/invoke.texi: Same.
27559 * doc/loop.texi: Same.
27560 * doc/md.texi: Same.
27561 * doc/objc.texi: Same.
27562 * doc/rtl.texi: Same.
27563 * doc/tm.texi: Same.
27564 * doc/tm.texi.in: Same.
27565 * doc/trouble.texi: Same.
27567 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27570 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27571 (collect_checksum_attributes): Set it.
27572 (die_checksum_ordered): Use it.
27574 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27576 PR tree-optimization/79908
27577 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27578 change: For a VA_ARG whose LHS has been cast away, use
27579 force_gimple_operand to construct the side effects.
27581 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27583 PR translation/80001
27584 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27585 more amenable to translation.
27586 (oacc_loop_auto_partitions): Likewise.
27588 2017-03-21 Marek Polacek <polacek@redhat.com>
27589 Martin Sebor <msebor@redhat.com>
27591 PR tree-optimization/80109
27592 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27593 on INTEGRAL_TYPE_P.
27595 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27596 Segher Boessenkool <segher@kernel.crashing.org>
27599 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27600 check reg_used_between_p between insn and one of succ or succ2
27601 depending on if succ is artificial insn not inserted into insn
27604 2017-03-21 Martin Liska <mliska@suse.cz>
27606 PR gcov-profile/80081
27607 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27608 * doc/gcc.texi: Include gcov-dump stuff.
27609 * doc/gcov-dump.texi: New file.
27611 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27613 PR rtl-optimization/79150
27614 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27615 conditional jump, if the jump is the last insn of the loop.
27617 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27618 Richard Biener <rguenth@suse.de>
27620 PR tree-optimization/79908
27621 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27622 been cast away, use force_gimple_operand to construct the side
27625 2017-03-21 Martin Liska <mliska@suse.cz>
27627 PR libfortran/79956
27628 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27631 2017-03-21 Brad Spengler <spender@grsecurity.net>
27634 * plugin.c (htab_hash_plugin): New function.
27635 (add_new_plugin): Use it and adjust.
27636 (parse_plugin_arg_opt): Adjust.
27637 (init_one_plugin): Likewise.
27639 2017-03-21 Richard Biener <rguenther@suse.de>
27641 PR tree-optimization/80032
27642 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27643 if set force the cleanup to happen unconditionally.
27644 (gimplify_target_expr): Push inserted clobbers with force_uncond
27645 to avoid them being removed by control-dependent DCE.
27647 2017-03-21 Richard Biener <rguenther@suse.de>
27649 PR tree-optimization/80122
27650 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27651 va_arg_pack_len when the inlined call stmt requires pack
27653 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27655 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27658 * tsan.c (instrument_builtin_call): If the memory model argument
27659 is not a constant, assume it is valid.
27662 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27665 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27667 PR rtl-optimization/79910
27668 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27669 if its dest is used by an insn before I2 (other than the combined
27670 insns themselves, which are properly handled already).
27672 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27675 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27677 * combine.c (record_used_regs): New static function.
27678 (try_combine): Handle situations where there is an additional
27679 instruction between I2 and I3 which needs to have a LOG_LINK
27683 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27685 * combine.c (try_combine): Delete redundant i1 test. Call
27686 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27688 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27691 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27692 alternatives 13/14.
27694 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27696 PR tree-optimization/80054
27697 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27698 the optimization if a PHI or any of its arguments is not dominated
27699 by the candidate's basis. Use gphi* rather than gimple* as
27701 (replace_profitable_candidates): Clean up a gimple* variable that
27702 should be a gphi* variable.
27704 2017-03-20 Martin Sebor <msebor@redhat.com>
27707 * doc/extend.texi (attribute constructor): Document present limitation.
27709 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27712 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27713 __POWER9_VECTOR__ #ifdef control, change template definition to
27714 use Power9-specific built-in function.
27715 (vec_any_eq): Likewise.
27716 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27717 to control outcomes from this test.
27718 (vector_ae_<mode>p): For VEC_F modes, likewise.
27720 2017-03-20 Ian Lance Taylor <iant@google.com>
27722 * config/i386/i386.c (ix86_function_regparm): Save an extra
27723 register for -fsplit-stack with DECL_STATIC_CHAIN.
27725 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27728 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27729 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27731 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27733 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27735 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27738 2017-03-20 Marek Polacek <polacek@redhat.com>
27741 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27743 2017-03-20 Richard Biener <rguenther@suse.de>
27745 PR tree-optimization/80113
27746 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27747 allocate extra SSA name for PHI def.
27748 (add_close_phis_to_outer_loops): Likewise.
27749 (add_close_phis_to_merge_points): Likewise.
27750 (copy_loop_close_phi_args): Likewise.
27751 (copy_cond_phi_nodes): Likewise.
27753 2017-03-20 Martin Liska <mliska@suse.cz>
27755 PR middle-end/79753
27756 * tree-chkp.c (chkp_build_returned_bound): Do not build
27757 returned bounds for a LHS that's not a BOUNDED_P type.
27759 2017-03-20 Martin Liska <mliska@suse.cz>
27763 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27764 COMPLEX_CST and VECTOR_CST.
27766 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27769 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27770 target operand. A new splitter adds the clobber statement in case
27771 the target operand is dead anyway.
27773 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27775 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27776 to age-old versions of binutils and glibc.
27778 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27780 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27782 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27784 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27786 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27788 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27789 requirement for binutils 2.13.
27791 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27793 * combine.c (try_combine): Delete redundant i1 test. Call
27794 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27796 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27798 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27799 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27801 <riscv64-*-elf>: Re-arrange section
27802 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27803 <riscv32-*-linux>: Likewise.
27804 <riscv64-*-elf>: Likewise
27805 <riscv64-*-linux>: Likewise.
27807 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27810 * aarch64.opt(verbose-cost-dump): Fix typo.
27812 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27815 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27816 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27818 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27820 * reload.c (find_reloads): When reloading a nonoffsettable address,
27821 use RELOAD_OTHER for it and its address reloads.
27823 PR rtl-optimization/79910
27824 * combine.c (record_used_regs): New static function.
27825 (try_combine): Handle situations where there is an additional
27826 instruction between I2 and I3 which needs to have a LOG_LINK
27829 2017-03-17 Jeff Law <law@redhat.com>
27831 PR tree-optimization/71437
27832 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27833 conditional in the hash table first.
27834 (vrp_dom_walker::before_dom_children): Extract condition from
27835 ASSERT_EXPR. Record condition, its inverion and any implied
27836 conditions as well.
27838 2017-03-17 Marek Polacek <polacek@redhat.com>
27839 Markus Trippelsdorf <markus@trippelsdorf.de>
27841 PR tree-optimization/80079
27842 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27845 2017-03-17 Richard Biener <rguenther@suse.de>
27847 PR middle-end/80075
27848 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27849 Properly verify the LHS before the RHS possibly claims to be
27851 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27854 2017-03-17 Martin Jambor <mjambor@suse.cz>
27856 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27857 (List of -O2 options): Likewise.
27858 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27861 2017-03-17 Tom de Vries <tom@codesourcery.com>
27863 * gcov-dump.c (print_usage): Print bug_report_url.
27865 2017-03-17 Richard Biener <rguenther@suse.de>
27867 PR middle-end/80050
27868 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27869 (parser::peek): Likewise.
27871 2017-03-17 Richard Biener <rguenther@suse.de>
27873 PR tree-optimization/80048
27874 * sese.c (free_sese_info): Properly release rename_map and
27875 copied_bb_map elements.
27877 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27879 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27880 Add linked-list forward and backlinks. Insert on
27881 construction, remove on destruction.
27882 (class pass_store_merging): Add m_stores_head field.
27883 (pass_store_merging::terminate_and_process_all_chains):
27884 Iterate over m_stores_head list.
27885 (pass_store_merging::terminate_all_aliasing_chains):
27887 (pass_store_merging::execute): Check for debug stmts first.
27888 Push new chains onto the m_stores_head stack.
27890 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27893 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27894 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27895 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27897 2017-03-16 Jeff Law <law@redhat.com>
27899 PR tree-optimization/71437
27900 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27901 member function. Implementation moved into after_dom_children
27902 member function and into the threader's thread_outgoing_edges
27904 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27905 some code into new thread_outgoing_edges.
27906 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27907 definition. Simplify marker handling (do it here). Assume we always
27908 have the available expression and the const/copies tables.
27909 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27911 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27912 * tree-vrp.c (equiv_stack): No longer file scoped.
27913 (vrp_dom_walker): New class.
27914 (vrp_dom_walker::before_dom_children): New member function.
27915 (vrp_dom_walker::after_dom_children): Likewise.
27916 (identify_jump_threads): Setup domwalker. Use it rather than
27917 walking edges in a random order by hand. Simplify setup/finalization.
27918 (finalize_jump_threads): Remove.
27919 (vrp_finalize): Do not call identify_jump_threads here.
27920 (execute_vrp): Do it here instead and call thread_through_all_blocks
27923 PR tree-optimization/71437
27924 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27926 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27928 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27929 (dom_opt_dom_walker::thread_across_edge): Remove
27930 handle_dominating_asserts argument. All callers changed.
27931 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27932 changes. Remove calls to lhs_of_dominating_assert. Other
27933 uses of handle_dominating_asserts turn into unconditional code
27934 (simplify_control_stmt_condition_1): Likewise.
27935 (simplify_control_stmt_condition): Likewise.
27936 (thread_through_normal_block, thread_across_edge): Likewise.
27937 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27938 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27939 object if it is not an SSA_NAME.
27940 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27941 before calling into the VRP specific simplifiers.
27942 (identify_jump_threads): Remove handle_dominating_asserts
27945 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27948 * tree-diagnostic.c (default_tree_printer): No longer static.
27949 * tree-diagnostic.h (default_tree_printer): New prototype.
27951 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27953 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27954 Change ins into fmov.
27956 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27958 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27959 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27960 Use h_con constraint for operand 1.
27961 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27962 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27964 2017-03-15 Jeff Law <law@redhat.com>
27966 PR tree-optimization/71437
27967 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27968 (record_temporary_equivalences): Use it.
27970 PR tree-optimization/71437
27971 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27972 tree-ssa-scopedtables.
27973 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27974 (record_conditions, record_cond, vuse_eq): Likewise.
27975 (record_edge_info): Adjust to API tweak of record_conditions.
27976 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27977 (record_temporary_equivalences, optimize_stmt): Likewise.
27978 (eliminate_redundant_computations): Likewise.
27979 (record_equivalences_from_stmt): Likewise.
27980 * tree-ssa-scopedtables.c: Include options.h and params.h.
27981 (vuse_eq): New function, moved from tree-ssa-dom.c
27982 (build_and_record_new_cond): Likewise.
27983 (record_conditions): Likewise. Accept vector of conditions rather
27984 than edge_equivalence structure for first argument.
27985 for the first argument.
27986 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27987 from tree-ssa-dom.c.
27988 (avail_exprs_stack::record_cond): Likewise.
27989 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27990 from tree-ssa-dom.c.
27991 (avail_exprs_stack): Add new member functions lookup_avail_expr
27993 (record_conditions): Declare.
27995 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27998 * lra-constraints.c (process_alt_operands): Increase reject for
27999 reloading an input/output operand.
28001 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28004 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28005 insns to convert from signed/unsigned char/short to IEEE 128-bit
28007 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28009 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28012 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28013 subreg of inner mode for values already in registers.
28015 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28017 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28018 iteration reg is used after the loop.
28020 2017-03-14 Martin Sebor <msebor@redhat.com>
28022 PR tree-optimization/79800
28023 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28024 precision in negative-positive range.
28025 (format_floating): Call non-const overload with adjusted precision.
28027 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28030 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28033 2017-03-14 Martin Sebor <msebor@redhat.com>
28035 PR middle-end/80020
28036 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28037 * builtins.def (aligned_alloc): Use it.
28040 * Makefile.in (GTFILES): Add calls.c.
28041 * calls.c: Include "gt-calls.h".
28043 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28045 PR rtl-optimization/79728
28046 * regs.h (struct target_regs): New field
28047 x_contains_allocatable_regs_of_mode.
28048 (contains_allocatable_regs_of_mode): New macro.
28049 * reginfo.c (init_reg_sets_1): Initialize it, and change
28050 contains_reg_of_mode so it includes global regs as well.
28051 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28052 rather than contains_regs_of_mode.
28054 2017-03-14 Martin Liska <mliska@suse.cz>
28056 * doc/invoke.texi: Document options that can't be combined with
28057 -fcheck-pointer-bounds.
28059 2017-03-14 Martin Liska <mliska@suse.cz>
28061 PR middle-end/79831
28062 * doc/invoke.texi (-Wchkp): Document the option.
28064 2017-03-14 Martin Liska <mliska@suse.cz>
28066 * Makefile.in: Install gcov-dump.
28068 2017-03-14 Martin Liska <mliska@suse.cz>
28070 * multiple_target.c (expand_target_clones): Bail out for
28071 an invalid attribute.
28073 2017-03-14 Richard Biener <rguenther@suse.de>
28075 * alias.c (struct alias_set_entry): Pack properly.
28076 * cfgloop.h (struct loop): Likewise.
28077 * cse.c (struct set): Likewise.
28078 * ipa-utils.c (struct searchc_env): Likewise.
28079 * loop-invariant.c (struct invariant): Likewise.
28080 * lra-remat.c (struct cand): Likewise.
28081 * recog.c (struct change_t): Likewise.
28082 * rtl.h (struct address_info): Likewise.
28083 * symbol-summary.h (function_summary): Likewise.
28084 * tree-loop-distribution.c (struct partition): Likewise.
28085 * tree-object-size.c (struct object_size_info): Likewise.
28086 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28087 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28088 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28089 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28090 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28091 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28092 (struct _stmt_vec_info): Likewise.
28094 2017-03-14 Martin Liska <mliska@suse.cz>
28097 * multiple_target.c (create_dispatcher_calls): Check that
28098 a target can create a function dispatcher.
28100 2017-03-14 Martin Liska <mliska@suse.cz>
28103 * multiple_target.c (expand_target_clones): Drop local.local
28104 flag for default implementation.
28106 2017-03-14 Richard Biener <rguenther@suse.de>
28108 PR tree-optimization/80030
28109 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28111 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28113 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28114 gcc_fallthrough() instead of __attribute__((fallthrough));
28116 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28118 * doc/gcc.texi: Remove "up" link to (DIR).
28119 * doc/gccint.texi: Ditto.
28121 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28123 * doc/install.texi (Specific) <avr>: Remove reference to
28126 2017-03-13 Jeff Law <law@redhat.com>
28128 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28129 attribute rather than comments.
28131 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28132 match_scratch operand is highest.
28134 2017-03-13 Martin Liska <mliska@suse.cz>
28136 PR middle-end/78339
28137 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28138 is a CHKP clone, use original declaration.
28140 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28142 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28143 (arc_conditional_register_usage): Use a different allocation order
28144 when optimizing for size.
28145 * common/config/arc/arc-common.c (arc_option_optimization_table):
28146 Section anchors default on when optimizing for size.
28148 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28150 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28152 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28154 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28155 * config/arc/arc.md (cpu_facility): Add cd variant.
28156 (*movqi_insn): Add code density variant.
28157 (*movhi_insn): Likewise.
28158 (*movqi_insn): Likewise.
28159 (*addsi3_mixed): Likewise.
28160 (subsi3_insn): Likewise.
28162 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28164 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28166 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28168 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28169 expressions with MINUS and UNARY ops.
28171 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28174 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28176 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28177 between vec_select and vector argument.
28178 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28179 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28180 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28181 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28182 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28183 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28185 2017-03-13 Richard Biener <rguenther@suse.de>
28188 * params.def (vect-max-peeling-for-alignment): Fix typo.
28190 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28192 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28193 issue that only occurred with binutils below 2.18.
28195 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28197 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28198 refer to binutils 2.11/2.12 minimum.
28200 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28202 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28203 ftp.kernel.org and simplify binutils requirement.
28205 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28207 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28209 (Optimize Options): Ditto. Also remove redundancy.
28211 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28213 PR translation/79848
28214 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28216 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28217 to G_ to avoid double translation.
28219 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28221 PR translation/79923
28222 * auto-profile.c (get_combined_location): Convert leading
28223 character of diagnostics to lower case and remove trailing period.
28224 (read_profile): Likewise for various diagnostics.
28225 * config/arm/arm.c (arm_option_override): Remove trailing period
28226 from various diagnostics.
28227 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28228 (msp430_expand_delay_cycles): Likewise.
28230 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28233 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28234 full command-line argument, rather than just "str".
28235 (aarch64_validate_march): Likewise.
28236 (aarch64_validate_mtune): Likewise.
28238 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28240 PR rtl-optimization/78911
28241 * lra-assigns.c (must_not_spill_p): New function.
28242 (spill_for): Use it.
28244 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28246 PR tree-optimization/79981
28247 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28248 ATOMIC_COMPARE_EXCHANGE ifn result.
28249 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28250 IFN_ATOMIC_COMPARE_EXCHANGE.
28252 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28255 * opts.c (parse_sanitizer_options): Add missing question mark to
28256 "did you mean" message.
28258 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28260 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28262 (VMULEUH_UNS): Likewise.
28263 (VMULOUB_UNS): Likewise.
28264 (VMULOUH_UNS): Likewise.
28265 * config/rs6000/rs6000.c (builtin_function_type): Remove
28266 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28268 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28271 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28272 x with result of extra_parsing_for_operand_code_0.
28273 (function_reader::extra_parsing_for_operand_code_0): Convert
28274 return type from void to rtx, returning x. When reading
28275 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28276 larger size containing struct block_symbol.
28278 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28280 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28281 -mfloat128-hardware without -m64.
28283 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28286 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28287 entries to the case statement that marks unsigned arguments to
28288 overloaded functions.
28290 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28292 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28293 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28295 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28298 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28299 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28301 2017-03-10 Martin Liska <mliska@suse.cz>
28305 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28306 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28309 2017-03-10 Olivier Hainque <hainque@adacore.com>
28311 * tree-switch-conversion (array_value_type): Start by resetting
28312 candidate type to it's main variant.
28314 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28316 PR rtl-optimization/79909
28317 * combine.c (try_combine): Use simplify_replace_rtx on individual
28318 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28319 of the whole CALL_INSN_FUNCTION_USAGE.
28321 PR tree-optimization/79972
28322 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28323 get_range_info on SSA_NAMEs. Formatting fixes.
28325 2017-03-10 Richard Biener <rguenther@suse.de>
28326 Jakub Jelinek <jakub@redhat.com>
28328 PR tree-optimization/77975
28329 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28330 edge to be constant.
28331 (get_val_for): For constant x return it. Formatting fix.
28332 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28333 would use the same bases as the current one.
28335 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28337 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28338 instead of vec_select for V1TImode.
28339 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28341 (VSX_LE_128): Add V1TI to this mode iterator.
28342 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28343 (*vsx_le_perm_store_<mode>): Likewise.
28344 (pre-reload splitter for VSX stores): Likewise.
28345 (post-reload splitter for VSX stores): Likewise.
28346 (*vsx_xxpermdi2_le_<mode>): Likewise.
28347 (*vsx_lxvd2x2_le_<mode>): Likewise.
28348 (*vsx_stxvd2x2_le_<mode>): Likewise.
28350 2017-03-09 Michael Eager <eager@eagercon.com>
28352 Correct failures with --enable-checking=yes,rtl.
28354 * config/microblaze/microblaze.c (microblaze_expand_shift):
28355 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28356 test for const0_rtx.
28357 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28358 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28360 2017-03-09 Richard Biener <rguenther@suse.de>
28362 PR tree-optimization/79977
28363 * graphite-scop-detection.c (scop_detection::merge_sese):
28364 Handle the case of extra exits to blocks dominating the entry.
28366 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28368 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28371 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28373 PR rtl-optimization/79949
28374 * lra-constraints.c (process_alt_operands): Check memory when
28375 trying to predict a cycle. Print about the overall increase.
28377 2017-03-09 Richard Biener <rguenther@suse.de>
28379 PR middle-end/79971
28380 * gimple-expr.c (useless_type_conversion_p): Preserve
28381 TYPE_SATURATING for fixed-point types.
28383 2017-03-09 Richard Biener <rguenther@suse.de>
28386 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28387 alignment of BLKmode params.
28389 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28392 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28393 (VALL_NO_V2Q): Likewise.
28395 * config/aarch64/aarch64-simd.md
28396 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28398 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28399 VALL_NO_V2Q mode iterator.
28400 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28402 2017-03-09 Martin Liska <mliska@suse.cz>
28404 PR tree-optimization/79631
28405 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28406 tree_int_cst_sign_bit just for INTEGER constants.
28408 2017-03-09 Martin Liska <mliska@suse.cz>
28412 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28415 2017-03-09 Marek Polacek <polacek@redhat.com>
28418 * tree.c (inchash::add_expr): Handle TREE_VEC.
28420 2017-03-09 Martin Liska <mliska@suse.cz>
28423 (chkp_narrow_size_and_offset): New function.
28424 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28425 (void chkp_parse_bit_field_ref): New function.
28426 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28427 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28429 2017-03-09 Martin Liska <mliska@suse.cz>
28432 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28433 (chkp_find_bounds_1): Remove gcc_unreachable.
28435 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28438 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28439 BUILT_IN_SYNC*, determine the access type from the size suffix and
28440 always build a MEM_REF with that type. Handle forgotten
28441 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28444 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28445 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28446 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28447 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28448 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28449 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28450 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28451 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28452 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28453 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28454 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28455 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28456 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28457 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28458 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28459 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28460 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28461 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28462 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28463 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28464 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28465 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28466 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28467 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28468 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28469 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28470 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28471 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28472 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28473 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28474 definitions outside of __OPTIMIZE__ guarded section.
28477 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28478 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28479 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28480 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28483 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28485 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28486 ("vfenez<mode>"): Add missing constraints.
28488 2017-03-08 Martin Sebor <msebor@redhat.com>
28491 * config/nds32/nds32.c (nds32_option_override):
28492 Fix misspelled diagnostic.
28494 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28497 * gimplify.c (gimplify_omp_for): Replace index var in outer
28498 taskloop statement with an artificial variable and add
28499 OMP_CLAUSE_PRIVATE clause for it.
28501 2017-03-08 Richard Biener <rguenther@suse.de>
28503 PR tree-optimization/79955
28504 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28505 for accesses that are completely outside of the variable.
28507 2017-03-08 Andrew Haley <aph@redhat.com>
28509 PR tree-optimization/79943
28510 * tree-ssa-loop-split.c (compute_new_first_bound): When
28511 calculating the new upper bound, (END-BEG) should be added, not
28514 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28516 * config/avr/avr.md (setmemhi): Make sure match_dup
28517 operand number comes before match_scratch.
28519 2017-03-08 Richard Biener <rguenther@suse.de>
28521 PR tree-optimization/79920
28522 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28523 with ncopies == 1 to ...
28524 (vect_transform_slp_perm_load): ... here. Properly compute
28525 all element loads by iterating VF times over the group. Do
28526 not handle ncopies (computed in a broken way) in
28527 vect_create_mask_and_perm.
28529 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28532 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28533 is a uniform vector, use uniform_vector_p return value instead of
28534 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28536 2017-03-07 Marek Polacek <polacek@redhat.com>
28538 PR middle-end/79809
28539 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28540 (alloca_call_type): Likewise.
28542 2017-03-07 Martin Liska <mliska@suse.cz>
28544 * gcov.c (process_args): Put comment to correct location.
28546 2017-03-07 Martin Liska <mliska@suse.cz>
28548 PR middle-end/68270
28549 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28550 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28551 (chkp_narrow_bounds_for_field): Likewise.
28552 (chkp_parse_array_and_component_ref): Pass one more argument to
28555 2017-03-07 Richard Biener <rguenther@suse.de>
28557 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28560 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28562 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28563 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28565 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28568 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28569 to end of description.
28570 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28572 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28574 PR rtl-optimization/79901
28575 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28577 (*avx512f_<code><mode>3<mask_name>): ... this.
28578 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28579 iterator instead of VI8_AVX2_AVX512BW.
28581 PR rtl-optimization/79901
28582 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28583 min/max expander, expand it using expand_vec_cond_expr.
28586 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28589 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28592 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28593 to void * for PCH reasons.
28594 * dwarf2out.c (output_loc_operands, output_die): Cast
28595 v.val_vec.array to unsigned char *.
28597 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28600 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28603 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28605 PR rtl-optimization/79571
28606 * lra-constraints.c (process_alt_operands): Calculate static
28607 reject and subtract it from overall when only addresses will be
28610 2017-03-06 Julia Koval <julia.koval@intel.com>
28613 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28614 incoming stack boundary to 128 for 64-bit targets.
28616 2017-03-06 Richard Biener <rguenther@suse.de>
28618 PR tree-optimization/79894
28619 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28620 to NULL after folding it.
28622 2017-03-06 Richard Biener <rguenther@suse.de>
28624 PR tree-optimization/79824
28625 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28626 check disabling peeling for gaps.
28628 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28630 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28631 attributes): Document gettimeofday.
28633 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28635 * config/s390/s390.c (s390_option_override_internal): Set
28636 PARAM_MIN_VECT_LOOP_BOUND
28638 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28640 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28641 * config/s390/s390.md: Likewise.
28643 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28646 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28647 (<avx2_avx512>_perm<mode>): Rename to ...
28648 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28650 (<avx512>_perm<mode>_mask): Rename to ...
28651 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28653 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28654 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28655 instead of VI8F_256_512.
28656 (avx512f_perm<mode>): New define_expand.
28657 (avx512f_perm<mode>_mask): Likewise.
28658 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28659 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28661 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28663 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28664 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28666 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28668 2017-03-06 Martin Liska <mliska@suse.cz>
28671 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28672 when having a SSA NAME w/o VAR_DECL assigned to it.
28674 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28676 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28677 msa_dpsub_<su>_d): Fix MODE for vec_select.
28679 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28681 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28683 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28685 2017-03-06 Richard Biener <rguenther@suse.de>
28687 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28688 * plugin.c (register_plugin_info): Likewise.
28689 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28691 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28693 * config/i386/sse.md (sse_storehps, sse_storelps,
28694 avx_<castmode><avxsizesuffix>_<castmode>,
28695 avx512f_<castmode><avxsizesuffix>_<castmode>,
28696 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28697 in condition that at least one operand is not a MEM.
28699 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28701 PR middle-end/79805
28702 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28703 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28705 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28706 gimple_call_nothrow_p flag based on whether original builtin can throw.
28707 If it can, emit following stmts on the fallthrough edge.
28708 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28709 don't create new bb if inserting just debug stmts on the edge, try to
28710 insert them on the fallthru bb or just reset debug stmts.
28712 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28715 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28716 restore recog_data (including the operand rtxes inside it) around
28717 the call to get_insn_template.
28719 2017-03-03 Martin Sebor <msebor@redhat.com>
28721 PR tree-optimization/79699
28722 * context.c (context::~context): Free MPFR caches to avoid
28723 a memory leak on program exit.
28725 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28727 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28728 Use wide_int::ulow () instead of .elt (0).
28730 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28732 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28733 (*pushxf): Limit oF constraint to 32bit targets and add oC
28734 constraint for 64bit targets.
28735 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28736 (*pushdf): Change rmF constraint to rmC.
28738 2017-03-03 Martin Liska <mliska@suse.cz>
28740 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28741 Remove unused variable.
28743 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28746 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28747 is a memory operand, increase num_memory.
28748 (ix86_expand_args_builtin): Likewise.
28750 2017-03-03 Jan Hubicka <jh@suse.cz>
28753 * ipa-devirt.c (maybe_record_node): Properly handle
28754 __cxa_pure_virtual visibility.
28756 2017-03-03 Martin Liska <mliska@suse.cz>
28758 PR tree-optimization/79803
28759 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28761 (pass_loop_prefetch::execute): Disabled optimization if an
28762 assumption about L1 cache size is not met.
28764 2017-03-03 Martin Liska <mliska@suse.cz>
28766 PR rtl-optimization/79574
28767 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28768 (hash_scan_set): Likewise.
28769 (dump_hash_table): Likewise.
28770 (hoist_code): Likewise.
28772 2017-03-03 Richard Biener <rguenther@suse.de>
28774 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28775 in place of uhigh ().
28776 (fixed_convert_from_real): Likewise.
28778 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28781 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28783 2017-03-03 Richard Biener <rguenther@suse.de>
28785 PR middle-end/79818
28786 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28787 TYPE_OVERFLOW_UNDEFINED check.
28789 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28791 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28793 (vector_ae_<mode>_p): Likewise.
28794 (vector_nez_<mode>_p): Likewise.
28795 (vector_ne_v2di_p): Likewise.
28796 (vector_ae_v2di_p): Likewise.
28797 (vector_ne_<mode>_p): Likewise.
28798 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28800 (vsx_tsqrt<mode>2_fe): Likewise.
28802 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28805 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28807 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28809 PR rtl-optimization/79780
28810 * cprop.c (one_cprop_pass): When second and further conditional trap
28811 in a single basic block is turned into an unconditional trap, turn it
28812 into a deleted note to avoid RTL verification failures.
28814 2017-03-02 Richard Biener <rguenther@suse.de>
28816 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28818 2017-03-02 Richard Biener <rguenther@suse.de>
28820 PR tree-optimization/79345
28822 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28823 param and abort the walk, returning -1 if it is hit.
28824 (walk_aliased_vdefs): Take a limit param and pass it on.
28825 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28826 defaulting to 0 and return a signed int.
28827 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28828 (check_defs): New helper.
28829 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28830 about uninitialized memory.
28831 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28832 bogus uninitialized warning.
28833 (fixed_convert_from_real): Likewise.
28835 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28837 PR tree-optimization/66768
28838 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28839 iv_use if base object can't be determined.
28841 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28843 PR tree-optimization/79345
28844 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28845 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28846 (get_pattern_stats): Initialize it.
28847 * genemit.c (gen_expand): Verify match_scratch numbers come after
28848 match_operand/match_dup numbers.
28849 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28850 match_scratch numbers.
28851 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28853 * config/s390/s390.md (trunctdsd2): Likewise.
28855 2017-03-02 Richard Biener <rguenther@suse.de>
28857 * wide-int.h (wide_int_storage::operator=): Implement in terms
28860 2017-03-02 Richard Biener <rguenther@suse.de>
28862 PR tree-optimization/79777
28863 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28864 the to insert expression to sth existing.
28866 2017-03-01 Martin Sebor <msebor@redhat.com>
28868 PR middle-end/79692
28869 * gimple-ssa-sprintf.c
28870 (directive::known_width_and_precision): New function.
28871 (format_integer): Use it.
28872 (get_mpfr_format_length): Consider the full range of precision
28873 when computing %g output with the # flag. Set the likely byte
28874 count to 3 rather than 1 when precision is indeterminate.
28875 (format_floating): Correct the lower bound of precision.
28877 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28879 * doc/invoke.texi: Document default code model for 64-bit Linux.
28881 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28884 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28885 udiv rather than div since input pattern is unsigned.
28887 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28889 * config/i386/i386.c (print_reg): Warn for values of
28890 unsupported size in integer register.
28892 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28895 * config/rs6000/predicates.md (current_file_function_operand): Do
28896 not allow self calls to be local if the function is replaceable.
28898 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28901 * config/rs6000/altivec.h (vec_ctz and others): Change the
28902 preprocessor macro that controls conditional compilation from
28903 _ARCH_PWR9 to __POWER9_VECTOR__.
28904 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28905 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28906 control (instead of _ARCH_PWR9 control) so that template
28907 definition uses power9-specific function.
28908 (vec_any_eq): Likewise.
28909 (vec_all_ne): Change macro definition to use a power9-specific
28910 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28911 _ARCH_PWR9 control).
28912 (vec_any_eq) Likewise.
28913 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28914 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28915 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28916 support for xvcmpnedp instruction.
28917 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28918 macro expansion so that Power9 implementation of vec_all_ne does
28919 not use the AltiVec predicate framework.
28920 (VCMPNEH_P): Likewise.
28921 (VCMPNEW_P): Likewise.
28922 (VCMPNED_P): Likewise.
28923 (VCMPNEFP_P): Likewise.
28924 (VCMPNEDP_P): Likewise.
28925 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28926 implementation of vec_any_eq to not use AltiVec predicate
28928 (VCMPAEH_P): Likewise.
28929 (VCMPAEW_P): Likewise.
28930 (VCMPAED_P): Likewise.
28931 (VCMPAEFP_P): Likewise.
28932 (VCMPAEDP_P): Likewise.
28933 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28934 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28935 not use the AltiVec predicate framework.
28936 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28937 of vec_any_eq to not use AltiVec predicate framework.
28938 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28939 support for predefined __POWER9_VECTOR__ macro to indicate that
28940 Power9 instruction selection is enabled.
28941 (altivec_overloaded_builtins): Remove extraneous
28942 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28943 function argument types RS6000_BTI_bool_V16QI and
28944 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28945 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28946 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28947 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28948 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28949 Power9 for implementations of vec_cmpne. Change the signature for
28950 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28951 (representing vec_all_ne) to remove the previously described first
28952 argument of type RS6000_BTI_INTSI, as this was an artifact of
28953 reliance on the AltiVec predicate framework, which is no longer
28954 used in the implementation of these functions. Add
28955 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28956 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28957 since, unlike the AltiVec predicate framework implementation, we
28958 do not share function descriptors between vec_alle and vec_anyeq.
28959 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28960 set of modes that receive special treatment even when
28961 TARGET_P9_VECTOR is true. The special treatment emits code that
28962 does not depend on Power9 instructions.
28963 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28964 define_expand to not rely on AltiVec predicate framework.
28965 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28967 (vector_ne_v2di_p): Change this define_expand to not rely on
28968 AltiVec predicate framework.
28969 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28971 (vector_ne_<mode>_p): Change this define_expand to not rely on
28972 AltiVec predicate framework.
28973 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28975 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28976 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28977 define_insn pattern.
28978 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28979 define_insn pattern because the xvcmpne<VSs>. instruction is not
28981 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28982 instruction is not supported.
28984 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28986 * config/nvptx/nvptx.c: Include intl.h.
28988 2017-03-01 Martin Jambor <mjambor@suse.cz>
28991 * ipa-prop.h (ipa_bits): Removed field known.
28992 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28993 to pointers. Adjusted their comments to warn about their sharing.
28994 (ipcp_transformation_summary): Change bits to a vector of pointers.
28995 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28996 (ipa_get_ipa_bits_for_value): Declare.
28997 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28998 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28999 (ipa_bits_hash_table): Likewise.
29000 (ipa_vr_ggc_hash_traits): Likewise.
29001 (ipa_vr_hash_table): Likewise.
29002 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29003 being pointers and vr_known being removed.
29004 (ipa_set_jf_unknown): Likewise.
29005 (ipa_get_ipa_bits_for_value): New function.
29006 (ipa_set_jfunc_bits): Likewise.
29007 (ipa_get_value_range): New overloaded functions.
29008 (ipa_set_jfunc_vr): Likewise.
29009 (ipa_compute_jump_functions_for_edge): Use the above functions to
29010 construct bits and vr parts of jump functions.
29011 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29012 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29014 (ipcp_grow_transformations_if_necessary): Also allocate
29015 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29017 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29018 them. Fix too long lines.
29019 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29020 vr_known being removed.
29021 (ipa_read_jump_function): Use new setter functions to construct bits
29022 and vr parts of jump functions or set them to NULL.
29023 (write_ipcp_transformation_info): Adjust for bits being pointers.
29024 (read_ipcp_transformation_info): Likewise.
29025 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29027 Include gt-ipa-prop.h.
29028 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29030 (ipcp_store_bits_results): Likewise.
29031 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29032 Do not write to existing jump functions but use a temporary instead.
29034 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29037 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29038 attempt to use its first operand as BIT_FIELD_REF base.
29040 2017-03-01 Richard Biener <rguenther@suse.de>
29042 PR middle-end/79721
29043 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29044 interpolating formula in wrapping arithmetic.
29045 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29047 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29049 PR tree-optimization/79734
29050 * tree-vect-generic.c (expand_vector_condition): Optimize
29051 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29052 Handle VEC_COND_EXPR where comparison has different inner width from
29053 type's inner width.
29055 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29057 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29058 markup, and similar issues. Remove @opindex entries for things
29059 that aren't options. Add missing -mmpy-option entries.
29061 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29063 PR tree-optimization/79737
29064 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29065 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29066 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29067 instead of byte_size. Formatting fix.
29068 (shift_bytes_in_array_right): Formatting fix.
29070 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29073 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29074 condition on optimize for the leaf function test.
29076 2017-02-28 Martin Liska <mliska@suse.cz>
29079 * read-rtl-function.c (function_reader::handle_unknown_directive):
29080 Bail out when one uses -flto.
29082 2017-02-28 Martin Liska <mliska@suse.cz>
29084 * common.opt: Replace space with tabular for options of <number>
29086 * config/i386/i386.opt: Show <number> value for
29087 -mlarge-data-threshold.
29088 * opts.c (print_filtered_help): Do not display number in hexadecimal
29091 2017-02-28 Martin Liska <mliska@suse.cz>
29093 * common.opt: Fix --help=option -Q for options which are of
29096 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29098 * config/i386/i386.c (print_reg): Error out for values
29099 of 8-bit size in invalid integer register.
29101 2017-02-28 Martin Sebor <msebor@redhat.com>
29103 PR tree-optimization/79691
29104 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29106 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29109 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29110 gcc_unreachable with output_operand_lossage.
29112 2017-02-28 Richard Biener <rguenther@suse.de>
29114 PR tree-optimization/79740
29115 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29117 (visit_nary_op): Insert the nary into the hashtable if we
29118 pattern-matched sth.
29119 * tree-ssa-pre.c (eliminate_insert): Robustify.
29121 2017-02-28 Richard Biener <rguenther@suse.de>
29123 PR middle-end/79731
29124 * fold-const.c (decode_field_reference): Reject out-of-bound
29127 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29129 * config/i386/i386.c: Include intl.h.
29130 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29131 instead of just cond ? "..." : "...".
29132 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29133 * coverage.c (read_counts_file): Likewise.
29134 * omp-offload.c: Include intl.h.
29135 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29136 of just cond ? "..." : "...".
29137 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29138 of just cond ? "..." : "...".
29140 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29143 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29145 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29146 'tune for' CPU name.
29147 * config/arm/arm-cpu-data.h: Regenerated.
29149 2017-02-28 Richard Biener <rguenther@suse.de>
29151 PR tree-optimization/79732
29152 * tree-inline.c (expand_call_inline): Do not shadow var.
29154 2017-02-28 Richard Biener <rguenther@suse.de>
29156 PR tree-optimization/79723
29157 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29158 address-space properly.
29160 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29162 * doc/optinfo.texi (Optimization groups): Fix option used for
29164 * doc/invoke.texi (-fopt-info): Document "omp".
29165 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29166 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29167 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29168 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29169 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29171 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29173 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29174 OPTGROUP_OMP with "omp".
29176 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29179 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29180 for arithmetic shift of unsigned V2DI.
29182 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29184 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29185 arc/linux.h headers.
29186 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29187 (LINK_SPEC): Likewise.
29188 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29189 (EXTRA_SPECS): Likewise.
29190 (STARTFILE_SPEC): Likewise.
29191 (ENDFILE_SPEC): Likewise.
29192 (LIB_SPEC): Likewise.
29193 (TARGET_SDATA_DEFAULT): Likewise.
29194 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29195 (MULTILIB_DEFAULTS): Likewise.
29196 (DWARF2_UNWIND_INFO): Likewise.
29197 * config/arc/big.h: New file.
29198 * config/arc/elf.h: Likewise.
29199 * config/arc/linux.h: Likewise.
29200 * config/arc/t-uClibc: Remove.
29202 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29204 PR tree-optimization/77536
29205 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29206 (tree_transform_and_unroll_loop): Use above function to compute the
29207 estimated niter of unrolled loop and use it when scaling profile.
29208 Also use count info rather than frequency if it's non-zero.
29209 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29210 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29211 (vect_transform_loop): Call above function.
29213 2017-02-27 Richard Biener <rguenther@suse.de>
29215 PR tree-optimization/45397
29216 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29217 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29218 (visit_nary_op): Add pattern matching for CSEing sign-changed
29219 or truncated operations with wider ones.
29221 2017-02-27 Richard Biener <rguenther@suse.de>
29223 PR tree-optimization/79690
29224 * tree-vect-stmts.c (vectorizable_store): Use vector type
29225 built from the DR with address-space.
29227 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29229 * doc/invoke.texi (Optimize Options): Refine the description
29230 of asan-use-after-return.
29232 2017-02-25 Alan Modra <amodra@gmail.com>
29234 PR rtl-optimization/79584
29235 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29236 base, not ad->base_term, the reg within base. Remove assertion
29237 that ad->base == ad->base_term. Replace gen_int_mode using
29238 bogus mode with const0_rtx.
29240 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29242 PR middle-end/79396
29243 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29244 FMA_EXPR like tcc_binary or tcc_unary.
29246 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29249 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29251 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29252 (output_loc_operands): Handle DW_OP_call_ref and
29253 DW_OP_GNU_variable_value.
29254 (struct variable_value_struct): New type.
29255 (struct variable_value_hasher): Likewise.
29256 (variable_value_hash): New variable.
29257 (string_types): Remove.
29258 (copy_loc_descr): New function.
29259 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29260 (prepend_loc_descr_to_each): New function.
29261 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29262 instead of add_loc_descr_to_each if the first argument is single
29263 location list and the second has multiple.
29264 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29265 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29266 when looking for variable value which doesn't have other location info.
29267 (loc_list_from_tree): Formatting fix.
29268 (gen_array_type_die): Simplify DW_AT_string_length handling.
29269 (adjust_string_types): Remove.
29270 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29271 string_types. Call resolve_variable_values.
29272 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29273 (resolve_addr_in_expr): Likewise. Add A argument.
29274 (copy_deref_exprloc): Remove deref argument. Adjust for the
29275 original expression being DW_OP_GNU_variable_value with optionally
29276 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29277 optionally after it.
29278 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29279 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29280 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29281 (variable_value_hasher::hash, variable_value_hasher::equal): New
29283 (resolve_variable_value_in_expr, resolve_variable_value,
29284 resolve_variable_values, note_variable_value_in_expr,
29285 note_variable_value): New functions.
29286 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29289 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29292 * opts.h (handle_generated_option): Add GENERATED_P argument.
29293 * opts-common.c (handle_option): Adjust function comment.
29294 (handle_generated_option): Add GENERATED_P argument, pass it to
29296 (control_warning_option): Pass false to handle_generated_option
29298 * opts.c (maybe_default_option): Pass true to handle_generated_option
29300 * optc-gen.awk: Likewise.
29302 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29304 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29305 a REG, look at the REG it is a SUBREG of.
29306 (splitter for cmpeqsi_t): Ditto.
29308 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29310 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29311 the special USEs with the pattern of the insn, not the insn itself.
29313 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29316 * doc/invoke.texi: Document -mload-store-pairs.
29318 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29319 Sandra Loosemore <sandra@codesourcery.com>
29321 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29322 argument isn't a CONST_INT.
29323 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29324 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29325 (nios2_validate_compare): Bypass alternate compare logic if *op2
29326 is not a CONST_INT.
29327 (ldstwm_operation_p): Return false if first_base is not a REG or
29328 if first_offset is not a CONST_INT.
29330 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29332 * config/cris/cris.md: Use correct operand in a define_peephole2.
29334 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29336 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29338 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29340 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29341 this_insn if it is an INSN or JUMP_INSN.
29342 (force_offsettable): Look at base, not at addr.
29343 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29344 on things that aren't necessarily CONST_INTs.
29346 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29348 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29349 -mfpmath=sse is the default also for x86-32 targets with SSE2
29350 instruction set when @option{-ffast-math} is enabled
29352 2017-02-24 Jeff Law <law@redhat.com>
29354 PR rtl-optimizatoin/79286
29355 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29358 2017-02-24 Richard Biener <rguenther@suse.de>
29360 PR tree-optimization/79389
29361 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29364 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29366 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29367 function comment to reflect reality.
29368 (loop_exits_before_overflow): Fix typo in function description.
29370 2017-02-24 Richard Biener <rguenther@suse.de>
29372 PR tree-optimization/79389
29373 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29374 properly that a threading opportunity exists. Detect conditional
29375 copy/constant propagation opportunities.
29377 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29379 * config/visium/visium.md (type): Add trap.
29380 (b): New mode attribute.
29381 (*btst): Rename into...
29382 (*btst<mode>): ...this and adjust.
29383 (*cbranchsi4_btst_insn): Rename into...
29384 (*cbranch<mode>4_btst_insn): ...this and adjust.
29385 (trap): New define_insn.
29387 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29389 PR tree-optimization/79389
29390 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29391 (noce_reversed_cond_code): New function.
29392 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29393 reversed_comparison_code. Formatting fix.
29394 (noce_try_store_flag): Test rev_cond != NULL in addition to
29395 reversed_comparison_code.
29396 (noce_try_store_flag_constants): Likewise.
29397 (noce_try_store_flag_mask): Likewise.
29398 (noce_try_addcc): Use rev_cond if non-NULL instead of
29399 reversed_comparison_code.
29400 (noce_try_cmove_arith): Likewise. Formatting fixes.
29401 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29402 (noce_find_if_block): Initialize rev_cond.
29403 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29404 instead of false as last argument never attempt to reverse it
29407 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29409 PR tree-optimization/79663
29410 * tree-predcom.c (combine_chains): Process refs in reverse order
29411 only for ZERO length chains, and add explaining comment.
29413 2017-02-23 Jeff Law <law@redhat.com>
29415 PR tree-optimization/79578
29416 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29417 in call to operand_equal_p.
29419 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29422 * config/i386/cpuid.h: Fix another undefined behavior.
29424 2017-02-23 Richard Biener <rguenther@suse.de>
29426 PR tree-optimization/79683
29427 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29428 vector types for data-refs.
29430 2017-02-23 Martin Liska <mliska@suse.cz>
29432 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29434 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29436 PR middle-end/79665
29437 * internal-fn.c (get_range_pos_neg): Moved to ...
29438 * tree.c (get_range_pos_neg): ... here. No longer static.
29439 * tree.h (get_range_pos_neg): New prototype.
29440 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29441 are known to be in between 0 and signed maximum inclusive, try to
29442 expand both unsigned and signed divmod and use the cheaper one from
29445 2017-02-22 Jeff Law <law@redhat.com>
29447 PR tree-optimization/79578
29448 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29449 to compare base operands.
29451 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29454 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29455 gpc_reg_operand instead of fpr_reg_operand.
29457 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29459 * config/mips/mips.c (mips_return_in_memory): Force FP
29460 vector types to be returned in memory for o32 ABI.
29462 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29464 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29465 instead of DW_TAG_member for static data member declarations and don't
29466 set no_linkage_name for static inline data members.
29467 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29470 2017-02-22 Martin Liska <mliska@suse.cz>
29472 * doc/invoke.texi: Replace inequality signs with square brackets
29475 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29478 * lra-constraints.c (simplify_operand_subreg): Handle
29479 WORD_REGISTER_OPERATIONS targets.
29481 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29484 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29485 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29486 elimination by swapping fld*.
29488 2017-02-22 Richard Biener <rguenther@suse.de>
29490 PR tree-optimization/79673
29491 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29492 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29493 irrelevant address-space qualifiers and avoiding a
29494 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29496 2017-02-22 Richard Biener <rguenther@suse.de>
29498 PR tree-optimization/79666
29499 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29500 to not symbolically negate if that may introduce undefined
29503 2017-02-22 Martin Liska <mliska@suse.cz>
29506 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29507 * data-streamer-out.c (streamer_write_gcov_count_stream):
29509 * value-prof.c (stream_out_histogram_value): Make assert more
29510 precise based on type of counter.
29512 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29515 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29516 Use nonimmediate_operand instead of memory_operand for operand 1.
29517 (float-extend standard_x87sse_constant_load splitter): Ditto.
29519 2017-02-21 Jeff Law <law@redhat.com>
29521 PR tree-optimization/79621
29522 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29523 blocks with edges to themselves.
29525 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29528 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29529 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29530 Use gimple_call_builtin_p.
29533 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29534 on temporarily removed DEBUG_INSNs.
29536 PR tree-optimization/79649
29537 * tree-loop-distribution.c (classify_partition): Give up on
29538 non-generic address space loads/stores.
29540 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29542 * doc/loop.texi (Loop manipulation): Remove nonexistent
29543 tree_ssa_loop_version from the documentation.
29544 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29546 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29549 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29550 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29551 * config/rs6000/rs6000.c: Include except.h.
29552 (rs6000_expand_split_stack_prologue): Call
29553 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29555 2017-02-21 Martin Jambor <mjambor@suse.cz>
29558 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29559 have been analyzed.
29561 2017-02-21 Martin Jambor <mjambor@suse.cz>
29563 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29564 for backward compatibility only.
29565 * doc/invoke.texi (Option Summary): Remove all references to
29566 -fipa-cp-alignment.
29568 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29572 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29574 * lra-constraints.c (curr_insn_transform): Handle
29575 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29577 2017-02-21 Martin Liska <mliska@suse.cz>
29579 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29581 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29583 PR translation/79638
29584 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29586 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29589 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29590 (arm_function_ok_for_sibcall): Return false for an indirect call by
29591 descriptor if all the argument registers are used.
29592 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29593 alignment of the function.
29595 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29597 PR tree-optimization/61441
29598 * simplify-rtx.c (simplify_const_unary_operation): For
29599 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29600 the sNaN unmodified.
29602 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29604 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29605 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29606 instead of SYSTEM_HEADER_DIR.
29608 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29609 Martin Liška <mliska@suse.cz>
29611 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29612 Fix typos and grammar, use active voice, and clarify.
29614 2017-02-20 Marek Polacek <polacek@redhat.com>
29616 PR middle-end/79537
29617 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29620 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29622 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29625 * config/i386/i386.c (ix86_expand_builtin): Handle
29626 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29627 ix86_builtins_isa[fcode].isa as a requirement of those
29628 flags and any other flag in the bitmask.
29629 (ix86_init_mmx_sse_builtins): Use 0 instead of
29630 ~OPTION_MASK_ISA_64BIT as mask.
29631 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29632 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29633 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29634 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29636 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29639 * lra-constraints.c (split_reg): Check requested split mode
29640 is supported by the register.
29642 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29644 * lra-constraints.c (simplify_operand_subreg): Remove early
29647 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29650 * lra-constraints.c (curr_insn_transform): Tighten condition
29651 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29653 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29656 * lra-constraints.c (curr_insn_transform): Handle
29657 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29659 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29662 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29664 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29666 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29669 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29672 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29674 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29677 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29679 * final.c (last_columnnum, override_columnnum): New variables.
29680 (final_start_function): Set last_columnnum, pass it to begin_prologue
29681 hook and pass 0 to dwarf2out_begin_prologue.
29682 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29683 to source_line debug hook.
29684 (notice_source_line): Compute last_columnnum and for debug_column_info
29685 return true on column changes.
29686 * debug.h (struct gcc_debug_hooks): Add column argument to
29687 source_line and begin_prologue hooks.
29688 (debug_nothing_int_charstar_int_bool): Remove prototype.
29689 (debug_nothing_int_int_charstar,
29690 debug_nothing_int_int_charstar_int_bool): New prototypes.
29691 (dwarf2out_begin_prologue): Add column argument.
29692 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29693 begin_prologue hooks.
29694 (debug_nothing_int_charstar_int_bool): Remove.
29695 (debug_nothing_int_int_charstar,
29696 debug_nothing_int_int_charstar_int_bool): New functions.
29697 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29698 through to dwarf2out_source_line.
29699 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29700 (dwarf2out_source_line): Add column argument, emit it if requested.
29701 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29703 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29704 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29705 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29706 through to dwarf2out_begin_prologue.
29707 (vmsdbgout_source_line): Add column argument, pass it through to
29708 dwarf2out_source_line.
29709 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29710 dbxout_source_line caller.
29711 (dbxout_source_line): Add column argument.
29713 * common.opt (gno-column-info, gcolumn-info): New options.
29714 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29715 (check_die): Also test for multiple DW_AT_decl_column attributes.
29716 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29717 DW_AT_decl_column if requested.
29718 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29720 (gen_variable_die): Likewise.
29721 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29722 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29725 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29726 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29727 (ix86_handle_option): Handle OPT_m3dnowa.
29728 * doc/invoke.texi (-m3dnowa): Document.
29729 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29730 -m3dnowa instead of -m3dnow -march=athlon.
29733 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29734 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29736 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29739 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29740 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29741 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29742 generator for vsx_xxpermdi_<mode>_be.
29743 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29744 force big-endian semantics.
29745 (vsx_xxpermdi_<mode>_be): New define_expand with same
29746 implementation as previous version of vsx_xxpermdi_<mode>.
29748 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29750 PR tree-optimization/79327
29751 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29752 variable, its initialization and use.
29754 2017-02-17 Julia Koval <julia.koval@intel.com>
29756 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29757 (OPTION_MASK_ISA_PKU_UNSET): New.
29758 (ix86_handle_option): Handle -mrdpid.
29759 * config/i386/cpuid.h (bit_RDPID): New.
29760 * config/i386/driver-i386.c (host_detect_local_cpu):
29761 Detect RDPID feature.
29762 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29763 * config/i386/i386-c.c (ix86_target_macros_internal):
29765 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29766 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29767 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29768 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29769 * config/i386/i386.md (define_insn "rdpid"): New.
29770 * config/i386/i386.opt Add -mrdpid.
29771 * config/i386/immintrin.h (_rdpid_u32): New.
29773 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29775 PR rtl-optimization/79541
29776 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29777 instead of transforming it into USE.
29779 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29781 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29782 If HONOR_SNANS (SFmode) force the input to a register.
29783 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29784 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29785 an frsp or similar insn.
29787 2017-02-17 Martin Liska <mliska@suse.cz>
29789 PR rtl-optimization/79577
29790 * params.def (selsched-max-sched-times): Increase minimum to 1.
29792 2017-02-17 Martin Liska <mliska@suse.cz>
29794 PR rtl-optimization/79574
29795 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29797 2017-02-17 Martin Liska <mliska@suse.cz>
29799 PR tree-optimization/79529
29800 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29801 ssa_defined_default_def_p to handle cases which are implicitly
29803 * tree-ssa.c (ssa_defined_default_def_p): New function.
29804 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29805 which are implicitly defined.
29806 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29808 2017-02-17 Richard Biener <rguenther@suse.de>
29810 PR middle-end/79576
29811 * params.def (max-ssa-name-query-depth): Limit to 10.
29813 2017-02-17 Richard Biener <rguenther@suse.de>
29815 PR tree-optimization/79552
29816 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29819 2017-02-17 Richard Biener <rguenther@suse.de>
29822 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29824 2017-02-17 Marek Polacek <polacek@redhat.com>
29826 PR middle-end/79536
29827 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29828 (fold_negate_expr): New wrapper.
29830 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29832 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29833 Correct terminology and de-emphasize pre-standard behavior.
29835 2017-02-16 Alan Modra <amodra@gmail.com>
29837 PR rtl-optimization/79286
29838 * ira.c (def_dominates_uses): New function.
29839 (update_equiv_regs): Don't create an equivalence for insns that
29840 may trap where the register def does not dominate the use.
29842 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29844 PR rtl-optimization/78127
29845 * lra.c (lra): Call lra_eliminate before finish the loop after
29848 2017-02-16 Richard Biener <rguenther@suse.de>
29850 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29852 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29853 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29854 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29855 (isl_val_int_from_wi): New function.
29856 (extract_affine_gmp): Rename to ...
29857 (extract_affine_wi): ... this, take a widest_int.
29858 (extract_affine_int): Just wrap extract_affine_wi.
29859 (add_param_constraints): Use isl_val_int_from_wi.
29860 (add_loop_constraints): Likewise, and extract_affine_wi.
29862 2017-02-15 Jeff Law <law@redhat.com>
29864 PR middle-end/79521
29865 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29866 ira_init_register_move_cost_if_necessary.
29868 2017-02-15 Martin Sebor <msebor@redhat.com>
29870 PR middle-end/32003
29871 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29872 removed in a prior commit.
29874 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29876 PR tree-optimization/79347
29877 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29878 counters during peeling.
29880 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29882 * Makefile.in (site.exp): Remove "set ISLVER".
29884 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29887 * real.c (real_from_integer): Call real_convert even for decimal.
29889 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29892 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29894 2017-02-14 Andrew Pinski <apinski@cavium.com>
29896 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29897 cores and change the partno/implementer to be correct.
29898 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29899 the 'B" as the implementer.
29900 * config/aarch64/aarch64-tune.md: Regenerate.
29902 2017-02-14 Carl Love <cel@us.ibm.com>
29904 * config/rs6000/rs6000.c: Add case statement entry to make the
29905 xvcvuxdsp built-in argument unsigned.
29906 * config/rs6000/vsx.md: Fix the source and return operand types so they
29907 match the instruction definitions from the ISA document. Fix typo
29908 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29911 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29914 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29915 member early_clobber_alts.
29916 * lra-lives.c (reg_early_clobber_p): New.
29917 (process_bb_lives): Use it.
29918 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29919 (debug_operand_data): Initialize early_clobber_alts.
29920 (setup_operand_alternative): Set up early_clobber_alts.
29921 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29922 alternatives to new_insn_reg.
29923 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29925 (lra_update_insn_regno_info): Pass the new arg.
29927 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29929 PR middle-end/79505
29930 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29931 (new_oacc_loop_raw): Don't clear already cleared fields.
29934 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29935 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29936 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29938 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29941 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29943 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29946 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29947 the extra instruction to the right place to store 128-bit constant
29950 2017-02-14 Martin Sebor <msebor@redhat.com>
29952 PR middle-end/79448
29953 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29954 warning for strings of unknown length.
29956 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29958 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29960 2017-02-14 Jeff Law <law@redhat.com>
29963 * ira-costs.c (scan_one_insn): Initialize register move costs
29964 for pseudos seen in USE/CLOBBER insns.
29966 PR tree-optimization/79095
29967 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29968 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29969 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29970 if the operands are known to be not equal, then the resulting range
29972 (intersect_ranges): If the new range is ~[0,0] and the old range is
29973 wide, then prefer ~[0,0].
29974 * tree-vrp.c (overflow_comparison_p_1): New function.
29975 (overflow_comparison_p): New function.
29976 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29977 if NAME is used in an overflow test.
29978 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29979 overflow check that can be expressed as an equality test, then adjust
29980 ops to be that equality test.
29982 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29984 * config/s390/s390-builtin-types.def: Remove flags argument.
29985 * config/s390/s390.c (s390_init_builtins): Likewise.
29987 2017-02-14 Martin Liska <mliska@suse.cz>
29989 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29990 vector. Fix trailing white spaces.
29992 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29994 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29997 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29999 PR rtl-optimization/68664
30000 * config/arm/arm.c (arm_sched_can_speculate_insn):
30001 New function. Declare prototype.
30002 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30004 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30006 PR rtl-optimization/68664
30007 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30009 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30011 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30013 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30014 max skip bytes for function, loop and jump.
30016 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30018 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30019 ABS_EXPR for gimple dump.
30021 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30024 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30026 PR tree-optimization/79408
30027 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30028 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30029 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30030 also if rhs1 is INTEGER_CST.
30032 2017-02-14 Richard Biener <rguenther@suse.de>
30034 PR middle-end/79432
30035 * tree-into-ssa.c (insert_phi_nodes): When the function can
30036 have abnormal edges rewrite SSA names with broken use-def
30037 dominance out of SSA and register them for PHI insertion.
30039 2017-02-13 Martin Sebor <msebor@redhat.com>
30041 PR middle-end/79496
30042 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30043 clearing info.nowrite flag when snprintf size argument is a range.
30045 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30047 * cprop.c (cprop_jump): Add missing space in string literal.
30048 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30049 (get_constraint_for_component_ref): Likewise.
30050 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30051 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30052 * lra-constraints.c (process_alt_operands): Likewise.
30053 * ipa-inline.c (inline_small_functions): Likewise.
30054 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30055 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30056 * trans-mem.c (diagnose_tm_1_op): Likewise.
30057 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30058 (grid_parallel_clauses_gridifiable): Likewise.
30060 * config/nvptx/mkoffload.c (process): Add space in between
30063 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30064 "MOD4_SSE_REGS" and "ALL_REGS".
30066 * spellcheck.c (test_data): Add , in between "foo" and "food".
30068 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30071 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30072 boundary crossing check and subsequent code generation agree.
30074 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30076 * config/aarch64/aarch64.c (has_memory_op): Delete.
30077 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30080 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30082 PR rtl-optimization/79388
30083 PR rtl-optimization/79450
30084 * combine.c (distribute_notes): When removing TEM_INSN for which
30085 corresponding dest has last value recorded, invalidate that last
30088 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30090 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30091 of explicit '@'. Add missing assembly comment marker on branch costs
30094 2017-02-13 Nathan Sidwell <nathan@acm.org>
30096 * gengtype-lex.l (<in_struct>): Add '/'.
30098 2017-02-13 Martin Liska <mliska@suse.cz>
30101 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30103 2017-02-13 Richard Biener <rguenther@suse.de>
30105 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30107 * configure: Re-generate.
30108 * config.in: Likewise.
30109 * graphite-dependences.c: Simplify as if
30110 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30111 * graphite-isl-ast-to-gimple.c: Likewise.
30112 * graphite-optimize-isl.c: Likewise.
30113 * graphite-poly.c: Likewise.
30114 * graphite-sese-to-poly.c: Likewise.
30115 * graphite.h: Likewise.
30116 * toplev.c: Include isl/version.h and use isl_version () for
30117 printing the ISL version.
30118 * doc/install.texi: Update ISL requirement.
30120 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30122 * doc/standards.texi (Standards): Update reference to
30125 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30127 * doc/extend.texi (Named Address Spaces): sourceware.org now
30129 * doc/install.texi (Binaries): Ditto.
30132 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30134 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30135 standard terminology "stringize"/"stringizing" throughout.
30136 * doc/cppinternals.texi: Likewise.
30138 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30140 * doc/extend.texi: Fix some spelling mistakes and typos.
30141 * doc/invoke.texi: Likewise.
30143 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30146 * params.def (inline-min-speedup) Change from 10 to 8.
30148 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30150 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30153 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30156 * ipa-inline-analysis.c (get_minimal_bb): New function.
30157 (record_modified): Use it.
30158 (remap_edge_change_prob): Handle also ancestor functions.
30160 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30162 * doc/contrib.texi (Contributors): Remove broken link into
30163 the Mauve CVS repository.
30165 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30167 PR middle-end/79454
30168 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30169 result computation whenever lhs doesn't have vector mode, not
30170 just when it has BLKmode.
30172 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30174 * doc/makefile.texi (profiledbootstrap): Refer to the
30175 installation instructions only in textual form.
30177 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30180 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30182 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30184 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30185 (Specific): Update mingw-w64 reference.
30187 (Specific): Remove broken link to Renesas RX processor.
30189 2017-02-10 Richard Biener <rguenther@suse.de>
30191 * toplev.c (process_options): Do not mention obsolete graphite
30192 options when printing sorry message about missing graphite support.
30193 Mention -floop-nest-optimize.
30195 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30197 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30198 (vtst_p16): Likewise.
30199 (vtstq_p8): Likewise.
30200 (vtstq_p16): Likewise.
30202 (vtstq_p64): Likewise.
30203 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30204 (vset_lane_p64): New.
30205 (vsetq_lane_p64): New.
30207 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30209 PR tree-optimization/79411
30210 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30211 stmt operands are SSA_NAMEs used in abnormal phis.
30212 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30215 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30218 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30221 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30223 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30225 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30227 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30231 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30232 not arbitrary TREE_CONSTANT.
30235 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30236 "omp declare target link" attribute unless is_global_var.
30237 * omp-offload.c (find_link_var_op): Likewise.
30239 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30240 Chung-Lin Tang <cltang@codesourcery.com>
30242 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30244 (gimplify_adjust_omp_clauses): Don't delete TILE.
30245 (gimplify_omp_for): Deal with TILE.
30246 * internal-fn.c (expand_GOACC_TILE): New function.
30247 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30249 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30250 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30252 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30253 avoid DIV for outermost collapse var.
30254 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30255 Remove out of date comments, fix whitespace.
30256 * omp-general.c (omp_extract_for_data): Deal with tiling.
30257 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30258 adjust OLF_DIM_BASE value.
30259 (struct omp_for_data): Add tiling field.
30260 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30261 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30262 for auto loops. Remove default auto determining, moved to
30263 oacc_loop_fixed_partitions.
30264 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30265 stmts, add e_mask field.
30266 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30267 (oacc_thread_numbers): Use oacc_dim_call.
30268 (oacc_xform_tile): New.
30269 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30270 (finish_oacc_loop): Adjust for ifns vector.
30271 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30272 add case for GOACC_TILE fns.
30273 (oacc_loop_xform_loop): Delete.
30274 (oacc_loop_process): Iterate over call list directly, and add
30275 handling for GOACC_TILE fns.
30276 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30278 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30279 vector partitioning to outer loops. Assign 2 partitions to loops
30280 when available. Add TILE handling.
30281 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30282 (execite_oacc_device_lower): Process GOACC_TILE fns,
30283 ignore unknown specs.
30284 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30285 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30286 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30288 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30290 * configure.ac (ACX_BUGURL): Update.
30291 * configure: Regenerate.
30293 2017-02-09 Richard Biener <rguenther@suse.de>
30295 PR tree-optimization/69823
30296 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30297 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30299 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30301 * config/arc/arc-c.def: Add __NPS400__ definition.
30302 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30303 (TARGET_NPS400): Define.
30305 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30307 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30309 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30310 pointer, arch_info.
30311 (arc_cpu_types): Fill the arch_info field with a pointer into the
30312 arc_arch_types table.
30313 (arc_selected_cpu): Declare.
30314 * config/arc/arc.c (arc_selected_cpu): Make global.
30315 (arc_selected_arch): Delete.
30316 (arc_base_cpu): Delete.
30317 (arc_override_options): Remove references to deleted variables,
30318 update access to arch information.
30319 (ARC_OPT): Update access to arch information.
30320 (ARC_OPTX): Likewise.
30321 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30322 (TARGET_ARC600): Update access to arch information.
30323 (TARGET_ARC601): Likewise.
30324 (TARGET_ARC700): Likewise.
30325 (TARGET_EM): Likewise.
30326 (TARGET_HS): Likewise.
30327 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30330 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30333 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30334 condition/operands for integer GE/LE/GEU/LEU operations.
30336 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30338 PR translation/79397
30339 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30342 2017-02-08 Martin Jambor <mjambor@suse.cz>
30345 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30346 whether allocation happened.
30347 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30348 nothing was allocated.
30350 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30352 PR tree-optimization/79408
30353 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30354 constant, but SSA_NAME with a known integer range, use the minimum
30355 of that range instead of op1 to determine if modulo can be replaced
30356 with its first operand.
30358 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30360 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30362 2017-02-08 Richard Biener <rguenther@suse.de>
30364 PR tree-optimization/71824
30365 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30366 Check all loops contained in the merged region.
30368 2017-02-07 Andrew Pinski <apinski@cavium.com>
30370 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30372 2017-02-07 Andrew Pinski <apinski@cavium.com>
30374 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30375 (thunderxt88): Likewise.
30376 (thunderxt81): Disable LSE and change v8.1 to v8.
30377 (thunderxt83): Likewise.
30379 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30380 Richard Biener <rguenther@suse.de>
30382 PR middle-end/79399
30383 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30384 type from int to size_t.
30385 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30387 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30389 PR rtl-optimization/79386
30390 * cprop.c (bypass_conditional_jumps): Initialize
30391 bypass_last_basic_block already before splitting bbs after
30392 unconditional traps...
30393 (bypass_conditional_jumps): ... rather than here.
30396 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30397 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30398 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30399 fix -masm=intel patterns.
30401 2017-02-07 Richard Biener <rguenther@suse.de>
30403 PR tree-optimization/79256
30404 PR middle-end/79278
30405 * builtins.c (get_object_alignment_2): Use min_align_of_type
30406 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30407 and ADJUST_FIELD_ALIGN.
30409 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30411 * doc/tm.texi: Regenerate.
30412 * stor-layout.c (layout_decl): Adjust.
30413 (update_alignment_for_field): Likewise.
30414 (place_field): Likewise.
30415 (min_align_of_type): Likewise.
30416 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30417 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30418 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30419 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30420 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30421 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30422 * config/i386/i386.c (x86_field_alignment): Likewise.
30423 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30424 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30425 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30426 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30427 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30428 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30432 2017-01-30 Richard Biener <rguenther@suse.de>
30434 PR tree-optimization/79256
30435 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30436 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30439 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30441 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30442 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30443 builtins to SImode and emit a zero-extend, if necessary.
30445 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30447 * docs/invoke.texi (RISC-V Options): Alphabetize.
30449 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30451 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30454 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30456 * config/riscv/riscv.c: New file.
30457 * common/config/riscv/riscv-common.c: Likewise.
30458 * config.gcc: Likewise.
30459 * config/riscv/constraints.md: Likewise.
30460 * config/riscv/elf.h: Likewise.
30461 * config/riscv/generic.md: Likewise.
30462 * config/riscv/linux.h: Likewise.
30463 * config/riscv/multilib-generator: Likewise.
30464 * config/riscv/peephole.md: Likewise.
30465 * config/riscv/pic.md: Likewise.
30466 * config/riscv/predicates.md: Likewise.
30467 * config/riscv/riscv-builtins.c: Likewise.
30468 * config/riscv/riscv-c.c: Likewise.
30469 * config/riscv/riscv-ftypes.def: Likewise.
30470 * config/riscv/riscv-modes.def: Likewise.
30471 * config/riscv/riscv-opts.h: Likewise.
30472 * config/riscv/riscv-protos.h: Likewise.
30473 * config/riscv/riscv.h: Likewise.
30474 * config/riscv/riscv.md: Likewise.
30475 * config/riscv/riscv.opt: Likewise.
30476 * config/riscv/sync.md: Likewise.
30477 * config/riscv/t-elf-multilib: Likewise.
30478 * config/riscv/t-linux: Likewise.
30479 * config/riscv/t-linux-multilib: Likewise.
30480 * config/riscv/t-riscv: Likewise.
30481 * configure.ac: Likewise.
30482 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30483 Waterman as RISC-V maintainers.
30484 * doc/install.texi: Add RISC-V entries.
30485 * doc/invoke.texi: Add RISC-V options section.
30486 * doc/md.texi: Add RISC-V constraints section.
30487 * configure: Regenerated.
30489 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30492 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30493 false values to be constant vectors with all 0 or all 1 bits set.
30494 (vcondu<mode><mode>): Likewise.
30495 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30497 (fpmask_comparison_operator): Update comment.
30498 (vecint_comparison_operator): New predicate.
30499 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30500 vector conditionals when the true and false values are constant
30501 vectors with all 0 bits or all 1 bits set.
30503 2017-02-06 Martin Sebor <msebor@redhat.com>
30505 PR tree-optimization/79376
30506 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30508 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30510 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30511 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30512 to simplify split condition.
30514 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30516 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30517 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30520 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30522 PR rtl-optimization/68664
30523 * target.def (can_speculate_insn): New hook.
30524 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30525 * doc/tm.texi: Regenerate.
30526 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30527 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30528 (rs6000_sched_can_speculate_insn): New function.
30530 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30532 PR tree-optimization/79284
30533 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30534 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30535 vectorizable_mask_load_store, vectorizable_operation,
30536 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30537 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30538 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30539 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30540 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30541 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30542 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30543 is_gimple_assign (stmt). Replace another such test with
30544 is_gimple_assign (stmt).
30546 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30549 * config/avr/avr.c (rtl-iter.h): Include it.
30550 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30551 (avr_legitimate_combined_insn): ...and implementation.
30553 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30555 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30556 * config/s390/s390.c (s390_const_operand_ok)
30557 (s390_canonicalize_comparison, s390_extract_part)
30558 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30559 (s390_contiguous_bitmask_p, s390_rtx_costs)
30560 (legitimize_pic_address): Likewise.
30561 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30562 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30563 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30564 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30565 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30567 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30569 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30570 REGNO($0) == REGNO($1).
30572 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30574 * config/s390/linux.h(SIZE_TYPE): Add comment.
30576 2017-02-06 Julian Brown <julian@codesourcery.com>
30577 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30578 Virendra Pathak <virendra.pathak@broadcom.com>
30580 * config/aarch64/aarch64-cores.def: Change the scheduler
30582 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30583 * config/aarch64/thunderx2t99.md: New file.
30585 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30587 * doc/standards.texi (Go Language): Update link to language
30590 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30592 * tree-eh.c (lower_resx): Sanitize profile.
30593 (cleanup_empty_eh_move_lp): Likewise.
30595 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30598 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30600 * cfgloopmanip.h (loop_version): Update prototype.
30601 * modulo-sched.c (sms_schedule): Update call of loop_version.
30602 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30603 * tree-parloops.c (gen_parallel_loop): Likewise.
30604 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30605 * tree-ssa-loop-split.c (split_loop): Likewise.
30606 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30607 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30609 2017-02-05 Martin Liska <mliska@suse.cz>
30612 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30614 (print_operand_address): Initialize a struct to zero.
30616 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30618 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30619 garbage collector only in textual form.
30621 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30623 * doc/extend.texi (x86 specific memory model extensions for
30624 transactional memory): Simplify a phrase.
30626 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30629 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30630 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30631 (atomic_storedi_1): Likewise.
30633 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30635 PR tree-optimization/79338
30636 * tree-parloops.c (gather_scalar_reductions): Don't call
30637 vect_analyze_loop_form for loop->inner before destroying loop's
30640 2017-02-03 Martin Sebor <msebor@redhat.com>
30642 PR tree-optimization/79327
30643 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30644 when precision has resulted in leading zeros.
30645 (format_integer): Adjust the likely counter to assume an unknown
30646 argument that may be zero is non-zero.
30648 2017-02-03 Jason Merrill <jason@redhat.com>
30651 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30652 avoid copying non-taken branch.
30654 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30656 PR tree-optimization/79340
30657 * tree-vect-loop.c (vectorizable_reduction): Release
30658 vec_defs elements after safe_splicing them into other vectors.
30661 PR tree-optimization/79327
30662 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30663 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30665 (format_integer): Use wide_int_to_tree instead of build_int_cst
30666 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30667 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30668 of shortest and longest sequence.
30670 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30672 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30673 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30675 2017-02-03 Walter Lee <walt@tilera.com>
30678 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30679 after initial stackframe link reg save.
30680 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30682 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30685 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30686 wu for stxssp alternative.
30688 2017-02-03 Martin Sebor <msebor@redhat.com>
30690 PR tree-optimization/79352
30691 * gimple-fold.c (get_range_strlen): Add argument.
30692 (get_range_strlen): Change return type to bool.
30693 (get_maxval_strlen): Pass in a dummy argument.
30694 * gimple-fold.h (get_range_strlen): Change return type to bool.
30695 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30696 * tree.h (array_at_struct_end_p): Add argument.
30697 * tree.c (array_at_struct_end_p): Handle it.
30699 2017-02-03 Martin Liska <mliska@suse.cz>
30702 * multiple_target.c (create_dispatcher_calls): Redirect edge
30703 from a caller of a dispatcher.
30704 (expand_target_clones): Make the clones local.
30705 (ipa_target_clone): Do both target clones and resolvers.
30706 (ipa_dispatcher_calls): Remove the pass.
30707 (pass_dispatcher_calls::gate): Likewise.
30708 (make_pass_dispatcher_calls): Likewise.
30709 * passes.def (pass_target_clone): Put as very first IPA early
30712 2017-02-03 Martin Liska <mliska@suse.cz>
30714 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30715 in case of a function with ifunc attribute.
30717 2017-02-03 Martin Liska <mliska@suse.cz>
30719 * cgraph.c (cgraph_node::dump): Dump function version info.
30720 * symtab.c (symtab_node::dump_base): Add missing new line.
30722 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30724 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30725 (ifcombine_ifandif): Use it.
30727 2017-02-03 Martin Liska <mliska@suse.cz>
30729 * doc/invoke.texi: Document default value for
30730 use-after-scope-direct-emission-threshold.
30732 2017-02-03 Martin Liska <mliska@suse.cz>
30734 PR tree-optimization/79339
30735 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30736 (format_floating): Likewise.
30738 2017-02-03 Martin Liska <mliska@suse.cz>
30741 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30743 (ipa_node_params_t::remove): Likewise.
30744 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30745 initialization from removed ipa_node_params_t::insert.
30746 (ipa_node_params::~ipa_node_params): Move from removed
30747 ipa_node_params_t::release.
30748 * symbol-summary.h (symbol_summary::m_released): New member.
30749 Do not release a summary twice. Do not allow to call finalizer
30750 for types of a summary that live in GGC memory.
30752 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30754 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30757 2017-02-02 Martin Sebor <msebor@redhat.com>
30759 PR middle-end/79275
30760 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30761 (format_string): Tighten up the range of output for non-constant
30762 strings and correct the expected range for wide non-constant strings.
30764 2017-02-02 Martin Sebor <msebor@redhat.com>
30766 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30768 PR middle-end/32003
30769 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30771 (-fdump-tree-@var): Add to index and document how to come up
30772 with pass-specific option and dump file names.
30773 (-fdump-passes): Clarify where to look for output.
30775 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30777 PR middle-end/77445
30778 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30779 statistics of the analyzed path; allow threading for speed when
30780 any of BBs along the path are optimized for speed.
30782 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30784 PR middle-end/78468
30785 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30786 settings of the virtual registers.
30789 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30791 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30792 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30795 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30797 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30798 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30800 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30802 * config/s390/s390.md: Add missing comments with the expanded
30804 * config/s390/vector.md: Likewise.
30805 * config/s390/vx-builtins.md: Likewise.
30807 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30810 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30811 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30812 conditions on a single line.
30814 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30816 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30817 __S390_VX__ to __VX__.
30819 2017-02-01 Andrew Pinski <apinski@cavium.com>
30821 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30822 stmt_info to record_stmt_cost.
30823 (vect_get_known_peeling_cost): Pass stmt_info if known to
30825 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30826 cpu_vector_cost field into
30827 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30828 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30829 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30830 splitting of scalar_stmt_cost and vec_stmt_cost.
30831 (thunderx_vector_cost): Likewise.
30832 (cortexa57_vector_cost): LIkewise.
30833 (exynosm1_vector_cost): Likewise.
30834 (xgene1_vector_cost): Likewise.
30835 (thunderx2t99_vector_cost): Improve after the splitting of the two
30837 (aarch64_builtin_vectorization_cost): Update for the splitting of
30838 scalar_stmt_cost and vec_stmt_cost.
30840 2017-02-01 Torvald Riegel <triegel@redhat.com>
30841 Richard Henderson <rth@redhat.com>
30843 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30844 conditional on existance of a fast atomic load.
30845 * optabs-query.c (can_atomic_load_p): New function.
30846 * optabs-query.h (can_atomic_load_p): Declare it.
30847 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30848 no fast atomic load is available for the particular size of access.
30849 (expand_atomic_compare_and_swap): Likewise.
30850 (expand_atomic_load): Likewise.
30851 (expand_atomic_store): Likewise.
30852 (expand_atomic_fetch_op): Likewise.
30853 * testsuite/lib/target-supports.exp
30854 (check_effective_target_sync_int_128): Remove x86 because it provides
30855 no fast atomic load.
30856 (check_effective_target_sync_int_128_runtime): Likewise.
30858 2017-02-01 Richard Biener <rguenther@suse.de>
30860 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30861 (graphite_transform_loops): Provide opt-info for optimized nests.
30862 * tree-parloop.c (parallelize_loops): Provide opt-info for
30863 parallelized loops.
30865 2017-02-01 Richard Biener <rguenther@suse.de>
30867 PR middle-end/79315
30868 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30869 was not set before.
30871 2017-02-01 Richard Biener <rguenther@suse.de>
30873 PR tree-optimization/71824
30874 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30875 Verify the loops are valid in the merged SESE region.
30876 (scop_detection::can_represent_loop_1): Check analyzing the
30877 evolution of the number of iterations in the region succeeds.
30879 2017-01-31 Ian Lance Taylor <iant@golang.org>
30881 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30882 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30884 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30886 PR preprocessor/79210
30887 * input.c (get_substring_ranges_for_loc): Replace line_width
30888 assertion with error-handling.
30890 2017-01-31 Richard Biener <rguenther@suse.de>
30892 PR tree-optimization/77318
30893 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30894 (create_pw_aff_from_tree): Take loop parameter.
30895 (add_condition_to_pbb): Pass loop of the condition to
30896 create_pw_aff_from_tree.
30898 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30900 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30901 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30903 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30907 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30909 (convert_int_to_float128): Likewise.
30910 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30911 (convert_int_to_float128): Likewise.
30912 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30913 (UNSPEC_IEEE128_CONVERT): Likewise.
30914 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30915 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30916 Use local variables for IBM extended format.
30917 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30918 (fix_trunc<mode>si2_fprs): Likewise.
30919 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30920 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30921 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30922 to know that we can now have integers of all sizes in vector
30924 (fix<uns>_<mode>di2_hw): Likewise.
30925 (float<uns>_<mode>si2_hw): Likewise.
30926 (fix_<mode>si2_hw): Likewise.
30927 (fixuns_<mode>si2_hw): Likewise.
30928 (float<uns>_<mode>di2_hw): Likewise.
30929 (float_<mode>di2_hw): Likewise.
30930 (float_<mode>si2_hw): Likewise.
30931 (floatuns_<mode>di2_hw): Likewise.
30932 (floatuns_<mode>si2_hw): Likewise.
30933 (xscvqp<su>wz_<mode>): Delete, no longer used.
30934 (xscvqp<su>dz_<mode>): Likewise.
30935 (xscv<su>dqp_<mode>): Likewise.
30936 (ieee128_mfvsrd_64bit): Likewise.
30937 (ieee128_mfvsrd_32bit): Likewise.
30938 (ieee128_mfvsrwz): Likewise.
30939 (ieee128_mtvsrw): Likewise.
30940 (ieee128_mtvsrd_64bit): Likewise.
30941 (ieee128_mtvsrd_32bit): Likewise.
30943 2017-01-31 Martin Liska <mliska@suse.cz>
30946 * ipa-prop.c (ipa_free_all_node_params): Call release method
30947 instead of ~sumbol_summary to not to trigger double times
30950 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30952 PR tree-optimization/71691
30953 * bitmap.h (class auto_bitmap): New.
30954 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30955 is_maybe_undefined instead of ssa_undefined_value_p.
30957 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30959 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30960 __S390_ARCH_LEVEL__ to __ARCH__.
30962 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30964 PR tree-optimization/79267
30965 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30966 if should_remove_lhs_p is true.
30968 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30971 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30972 (add_alignment_attribute): New.
30973 (base_type_die): Add alignment attribute.
30974 (subrange_type_die): Likewise.
30975 (modified_type_die): Likewise.
30976 (gen_array_type_die): Likewise.
30977 (gen_descr_array_type_die: Likewise.
30978 (gen_enumeration_type_die): Likewise.
30979 (gen_subprogram_die): Likewise.
30980 (gen_variable_die): Likewise.
30981 (gen_field_die): Likewise.
30982 (gen_ptr_to_mbr_type_die): Likewise.
30983 (gen_struct_or_union_type_die): Likewise.
30984 (gen_subroutine_type_die): Likewise.
30985 (gen_typedef_die): Likewise.
30986 (base_type_cmp): Compare alignment attribute.
30988 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30991 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30992 (setb_unsigned) New pattern for setb with CCUNS.
30993 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30994 subfc./subfe sequence to avoid overflow problems. Generate a
30995 shorter sequence with cmpld/setb for power9.
30996 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30997 for generating subfc. instruction.
30998 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30999 now uses this instruction.
31001 2017-01-30 Ian Lance Taylor <iant@google.com>
31004 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31005 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31007 2017-01-30 Martin Sebor <msebor@redhat.com>
31009 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31010 Move constant to the right of a relational operator.
31011 (get_mpfr_format_length, format_character, format_string): Ditto.
31012 (should_warn_p, maybe_warn): Same.
31014 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31016 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31019 * asan.c (get_translation_unit_decl): Remove function.
31020 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31022 2017-01-30 Martin Liska <mliska@suse.cz>
31024 PR gcov-profile/79259
31025 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31026 -fprofile-generate.
31028 2017-01-30 Martin Liska <mliska@suse.cz>
31031 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31032 Initialize variables with NULL value.
31034 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31037 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31039 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31041 2017-01-30 Richard Biener <rguenther@suse.de>
31043 PR tree-optimization/79276
31044 * tree-vrp.c (process_assert_insertions): Properly adjust common
31045 when removing a duplicate.
31047 2017-01-30 Richard Biener <rguenther@suse.de>
31049 PR tree-optimization/79256
31050 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31051 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31053 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31055 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31058 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31059 ("*r<noxa>sbg_<mode>_sll_bitmask")
31060 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31061 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31062 Use contiguous_bitmask_nowrap_operand.
31064 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31067 * config/rs6000/altivec.h (vec_xl): Revise #define.
31068 (vec_xst): Likewise.
31070 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31072 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31074 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31076 PR rtl-optimization/79194
31077 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31078 traps before call to bypass_conditional_jumps.
31080 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31082 PR tree-optimization/71374
31083 * lra-constraints.c (check_conflict_input_operands): New.
31084 (match_reload): Use it.
31086 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31089 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31090 account to calculate conflict_set.
31092 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31094 PR rtl-optimization/78559
31095 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31096 other_insn in combine.
31098 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31100 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31101 uint16_type_node for BT_UINT16.
31103 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31105 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31106 "RTL Tests" to menu.
31107 (GIMPLE Tests): New node.
31108 (RTL Tests): New node.
31110 2017-01-27 Richard Biener <rguenther@suse.de>
31112 PR tree-optimization/79245
31113 * tree-loop-distribution.c (distribute_loop): Apply cost
31114 modeling also to detected patterns.
31116 2017-01-27 Richard Biener <rguenther@suse.de>
31118 PR tree-optimization/71433
31119 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31120 (compare_assert_loc): New function.
31121 (process_assert_insertions): Sort and optimize assert locations
31122 to remove duplicates and push down identical assertions on
31123 edges to their destination block.
31125 2017-01-27 Richard Biener <rguenther@suse.de>
31127 PR tree-optimization/79244
31128 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31129 out SSA names even if abnormal.
31131 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31133 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31134 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31135 instead of MPFR_RNDN.
31137 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31140 * arm.c (arm_option_override): Don't call build_target_option_node
31141 until after doing all option overrides.
31142 (arm_valid_target_attribute_tree): Likewise.
31144 2017-01-27 Martin Liska <mliska@suse.cz>
31146 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31147 for {cd}tors and C++ {cd}tors.
31149 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31151 * config/s390/s390.md ("*setmem_long_and")
31152 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31154 2017-01-26 Martin Sebor <msebor@redhat.com>
31156 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31159 2017-01-26 Martin Sebor <msebor@redhat.com>
31161 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31162 HAVE_DFmode before using XFmode or DFmode.
31163 (parse_directive): Avoid using the z length modifier to avoid
31164 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31166 PR middle-end/78703
31167 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31168 to accept adjustment as an array.
31169 (get_int_range): New function.
31170 (struct directive): Make width and prec arrays.
31171 (directive::set_width, directive::set_precision): Call get_int_range.
31172 (format_integer, format_floating): Handle width and precision ranges.
31173 (format_string, parse_directive): Same.
31175 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31178 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31179 comdat_type_p set, just clone them, but keep the children in the
31183 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31184 which have direct callers with -fvar-tracking-assignments enabled
31186 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31187 inside of type units.
31189 2017-01-26 Martin Sebor <msebor@redhat.com>
31191 PR middle-end/78703
31192 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31194 (struct format_result): Replace number_chars, number_chars_min,
31195 and number_chars_max with a single member of struct result_range.
31197 (format_result::operator+=): Adjust.
31198 (struct fmtresult): Remove bounded. Handle likely and unlikely
31200 (fmtresult::adjust_for_width_or_precision): New function.
31201 (fmtresult:type_max_digits): New function.
31202 (bytes_remaining): Handle likely and unlikely counters.
31203 (min_bytes_remaining): Remove.
31204 (format_percent): Simplify.
31205 (format_integer, format_floating): Set likely and unlikely counters.
31206 (get_string_length, format_character, format_string): Same.
31207 (format_plain, should_warn_p): New function.
31208 (maybe_warn): Call should_warn_p. Update diagnostic messages
31209 and handle those for all directives, including plain strings.
31210 (format_directive): Handle likely and unlikely counters.
31211 Remove unnecessary quoting from diagnostics. Add an informational
31213 (add_bytes): Remove.
31214 (pass_sprintf_length::compute_format_length): Simplify.
31215 (try_substitute_return_value): Handle likely and unlikely counters.
31217 2017-01-26 Carl Love <cel@us.ibm.com>
31219 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31220 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31222 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31225 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31226 endianess for subregs into account.
31227 * lra-constraints.c (lra_constraints): Do risky transformations
31228 always on the first iteration.
31229 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31230 last_call_used_reg_set.
31231 (process_bb_lives): Define and use last_call_used_reg_set.
31232 * lra.c (lra): Always continue after lra_constraints on the first
31235 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31237 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31239 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31241 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31243 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31244 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31245 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31246 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31247 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31248 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31249 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31250 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31251 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31253 2017-01-26 Marek Polacek <polacek@redhat.com>
31256 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31257 for the third operand.
31259 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31261 PR middle-end/79236
31262 * omp-low.c (struct omp_context): Add simt_stmt field.
31263 (scan_omp_for): Return omp_context *.
31264 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31265 context to the _simt_ SIMD stmt.
31266 (lower_omp_for): For combined SIMD with sibling _simt_
31267 SIMD, make sure to use the same decls in _looptemp_
31268 clauses as in the sibling.
31270 2017-01-26 David Sherwood <david.sherwood@arm.com>
31272 PR middle-end/79212
31273 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31276 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31279 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31280 emit fld b; fld a; if possible.
31282 * brig-builtins.def: Update copyright years.
31283 * config/arm/arm_acle_builtins.def: Update copyright years.
31285 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31288 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31289 constraint instead of o for the stxsd instruction.
31291 2017-01-25 Carl Love <cel@us.ibm.com>
31293 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31294 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31296 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31298 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31300 2017-01-25 Richard Biener <rguenther@suse.de>
31302 PR tree-optimization/69264
31303 * target.def (vector_alignment_reachable): Improve documentation.
31304 * doc/tm.texi: Regenerate.
31305 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31307 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31308 earlier changes with respect to TYPE_USER_ALIGN.
31309 (vector_alignment_reachable_p): Likewise. Improve dumping.
31311 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31314 * config/arm/arm.md (xordi3): Force constant operand into a register
31317 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31319 * doc/invoke.texi (-fstore-merging): Correct default optimization
31320 levels at which it is enabled.
31321 (-O): Move -fstore-merging from list to...
31324 2017-01-25 Richard Biener <rguenther@suse.de>
31327 * omp-expand.c: Include debug.h.
31328 (expand_omp_taskreg): Make sure to generate early debug before
31329 outlining anything from a function.
31330 (expand_omp_target): Likewise.
31331 (grid_expand_target_grid_body): Likewise.
31333 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31336 * asan.c (get_translation_unit_decl): New function.
31337 (asan_add_global): Extract modules file name from globals
31338 TRANSLATION_UNIT_DECL name.
31340 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31343 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31344 for long calls with APCS frame and VFP.
31346 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31348 * cfg.c (original_copy_tables_initialized_p): New function.
31349 * cfg.h (original_copy_tables_initialized_p): New decl.
31350 * cfgrtl.c (relink_block_chain): Guard the call to
31351 free_original_copy_tables with a call to
31352 original_copy_tables_initialized_p.
31353 * cgraph.h (symtab_node::native_rtl_p): New decl.
31354 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31355 (symtab_node::needed_p): Don't assert for early assembly output
31356 for __RTL functions.
31357 (cgraph_node::finalize_function): Set "force_output" for __RTL
31359 (cgraph_node::analyze): Bail out early for __RTL functions.
31360 (analyze_functions): Update assertion to support __RTL functions.
31361 (cgraph_node::expand): Bail out early for __RTL functions.
31362 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31364 * function.h (struct function): Update comment for field
31366 * gimple-expr.c: Include "tree-pass.h".
31367 (gimple_has_body_p): Return false for __RTL functions.
31368 * Makefile.in (OBJS): Add run-rtl-passes.o.
31369 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31371 (gcc::pass_manager::get_clean_slate): New accessor.
31372 * passes.c: Include "insn-addr.h".
31373 (should_skip_pass_p): Add logging. Update logic for running
31374 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31375 property-provider override so it is only done for gimple passes.
31377 (skip_pass): New function.
31378 (execute_one_pass): Call skip_pass when skipping passes.
31379 * read-md.c (md_reader::read_char): Support filtering
31380 the input to a subset of line numbers.
31381 (md_reader::md_reader): Initialize fields
31382 m_first_line and m_last_line.
31383 (md_reader::read_file_fragment): New function.
31384 * read-md.h (md_reader::read_file_fragment): New decl.
31385 (md_reader::m_first_line): New field.
31386 (md_reader::m_last_line): New field.
31387 * read-rtl-function.c (function_reader::create_function): Only
31388 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31389 curr_properties. Set DECL_INITIAL to a dummy block.
31390 (read_rtl_function_body_from_file_range): New function.
31391 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31393 * run-rtl-passes.c: New file.
31394 * run-rtl-passes.h: New file.
31396 2017-01-24 Jeff Law <law@redhat.com>
31398 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31401 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31403 PR tree-optimization/79159
31404 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31405 (record_nonwrapping_iv): Improve boundary using above function if no
31406 value range information.
31408 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31409 Martin Jambor <mjambor@suse.cz>
31411 * brig-builtins.def: New file.
31412 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31413 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31414 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31415 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31416 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31417 * builtin-types.def (BT_INT8): New.
31418 (BT_INT16): Likewise.
31419 (BT_UINT8): Likewise.
31420 (BT_UINT16): Likewise.
31421 (BT_FN_ULONG): Likewise.
31422 (BT_FN_UINT_INT): Likewise.
31423 (BT_FN_UINT_ULONG): Likewise.
31424 (BT_FN_UINT_LONG): Likewise.
31425 (BT_FN_UINT_PTR): Likewise.
31426 (BT_FN_ULONG_PTR): Likewise.
31427 (BT_FN_INT8_FLOAT): Likewise.
31428 (BT_FN_INT16_FLOAT): Likewise.
31429 (BT_FN_UINT32_FLOAT): Likewise.
31430 (BT_FN_UINT16_FLOAT): Likewise.
31431 (BT_FN_UINT8_FLOAT): Likewise.
31432 (BT_FN_UINT64_FLOAT): Likewise.
31433 (BT_FN_UINT16_UINT32): Likewise.
31434 (BT_FN_UINT32_UINT16): Likewise.
31435 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31436 (BT_FN_INT_PTR_INT): Likewise.
31437 (BT_FN_UINT_PTR_UINT): Likewise.
31438 (BT_FN_LONG_PTR_LONG): Likewise.
31439 (BT_FN_ULONG_PTR_ULONG): Likewise.
31440 (BT_FN_VOID_UINT64_UINT64): Likewise.
31441 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31442 (BT_FN_INT8_INT8_INT8): Likewise.
31443 (BT_FN_INT16_INT16_INT16): Likewise.
31444 (BT_FN_INT_INT_INT): Likewise.
31445 (BT_FN_UINT_FLOAT_UINT): Likewise.
31446 (BT_FN_FLOAT_UINT_UINT): Likewise.
31447 (BT_FN_ULONG_UINT_UINT): Likewise.
31448 (BT_FN_ULONG_UINT_PTR): Likewise.
31449 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31450 (BT_FN_UINT_UINT_UINT): Likewise.
31451 (BT_FN_VOID_UINT_PTR): Likewise.
31452 (BT_FN_UINT_UINT_PTR: Likewise.
31453 (BT_FN_UINT32_UINT64_PTR): Likewise.
31454 (BT_FN_INT_INT_UINT_UINT): Likewise.
31455 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31456 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31457 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31458 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31459 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31460 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31461 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31462 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31463 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31464 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31465 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31466 * doc/frontends.texi: List BRIG FE.
31467 * doc/install.texi (Testing): Add BRIG tesring requirements.
31468 * doc/invoke.texi (Overall Options): Mention BRIG.
31469 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31471 2017-01-24 Richard Biener <rguenther@suse.de>
31473 PR translation/79208
31474 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31476 2017-01-24 Martin Jambor <mjambor@suse.cz>
31479 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31480 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31481 and known_contexts.
31483 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31485 PR middle-end/79123
31486 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31487 casts from signed to unsigned really don't have a range.
31489 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31491 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31492 GMP_RNDx for compatiblity.
31494 2017-01-24 Martin Liska <mliska@suse.cz>
31497 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31498 that would prevent us to call alloca with -1 as argument.
31500 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31502 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31503 Avoid -Wformat-security warning.
31505 2017-01-23 Andrew Pinski <apinski@cavium.com>
31507 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31510 2017-01-23 Martin Sebor <msebor@redhat.com>
31512 PR middle-end/78703
31513 * gimple-ssa-sprintf.c (warn_level): New global.
31514 (format_integer): Use it here and throughout the rest of the file.
31515 Use the same switch to compute sign as base.
31516 (maybe_warn): New function.
31517 (format_directive): Factor out warnings into maybe_warn.
31518 Add debugging output. Use warn_level.
31519 (add_bytes): Use warn_level.
31520 (pass_sprintf_length::compute_format_length): Add debugging output.
31521 (try_substitute_return_value): Same.
31522 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31524 PR middle-end/78703
31525 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31526 (struct fmtresult, format_integer, format_floating): Adjust.
31527 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31528 (get_string_length, format_string,format_directive): Same.
31529 (pass_sprintf_length::compute_format_length): Same.
31530 (try_substitute_return_value): Simplify slightly.
31532 PR middle-end/78703
31533 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31534 (fmtresult::operator+=): Outlined.
31535 (struct fmtresult): Add ctors.
31536 (struct conversion_spec): Rename...
31537 (struct directive): ...to this. Add and remove data members.
31538 (directive::set_width, directive::set_precision): New functions.
31539 (format_percent): Use fmtresult ctor.
31540 (get_width_and_precision): Remove.
31541 (format_integer): Make naming changes. Avoid computing width and
31543 (format_floating): Same. Adjust indentation.
31544 (format_character, format_none): New functions.
31545 (format_string): Moved character handling to format_character.
31546 (format_directive): Remove arguments, change return type.
31547 (parse_directive): New function.
31548 (pass_sprintf_length::compute_format_length): Move directive
31549 parsing to parse_directive.
31551 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31553 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31554 (assign_assembler_name_if_needed): ... this.
31555 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31556 (assign_assembler_name_if_needed): ... this.
31557 (free_lang_data_in_cgraph): Adjust callers.
31558 * cgraphunit.c (cgraph_node::analyze): Likewise.
31559 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31562 2017-01-23 Richard Biener <rguenther@suse.de>
31564 PR tree-optimization/79088
31565 PR tree-optimization/79188
31566 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31567 resetting loop bounds after last path deletion. Reset loop
31568 bounds of the target loop, make code match the comments.
31569 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31570 Make sure loops need no fixups.
31572 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31574 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31575 exponent support with double type for first argument.
31576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31577 type returned by __builtin_vec_extract_sig,
31578 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31579 functions from "vector int" to "vector unsigned int" or from
31580 "vector long long int" to "vector unsigned long long int".
31581 Changed type returned by __builtin_vec_extract_exp,
31582 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31583 functions from "vector int" to "vector unsigned int" or from
31584 "vector long long int" to "vector unsigned long long int".
31585 Changed return type of __builtin_vec_test_data_class,
31586 __builtin_vec_test_data_class_sp, and
31587 __builtin_vec_test_data_class_dp from "vector int" to
31588 "vector bool int" or from "vector long long int" to "vector bool
31589 long long int" and changed second argument type from "unsigned
31590 int" to "int". Added new overloaded function forms "vector float
31591 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31592 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31593 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31594 double, vector unsigned long long int)" and "vector double
31595 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31596 long int)". Changed return type of
31597 __builtin_scalar_test_data_class and
31598 __builtin_scalar_test_data_class_sp and
31599 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31600 int" and changed second argument from "unsigned int" to "int".
31601 Changed type returned by __builtin_scalar_test_neg,
31602 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31603 from "int" to "bool int". Added new overloaded function form
31604 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31605 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31606 exponent double-precision with floating point first argument.
31607 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31608 documentation of scalar_test_data_class, scalar_test_neg,
31609 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31610 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31611 vec_test_data_class built-in functions to reflect refinements in
31612 their type signatures.
31614 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31616 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31618 (aarch64_elf_asm_destructor): Likewise.
31620 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31622 PR rtl-optimization/78634
31623 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31624 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31625 * ifcvt.c (noce_try_cmove): Add missing cost check.
31627 PR rtl-optimization/71724
31628 * combine.c (if_then_else_cond): Look for situations where it is
31629 beneficial to undo the work of one of the recursive calls.
31631 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31633 PR tree-optimization/70754
31634 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31635 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31636 combined stmt before it if not NULL.
31637 (combine_chains): Process refs reversely and compute dominance point
31640 2017-01-23 Martin Liska <mliska@suse.cz>
31642 PR tree-optimization/79196
31643 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31644 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31646 (strlen_optimize_stmt): Call the renamed function.
31648 2017-01-23 Michael Matz <matz@suse.de>
31650 PR tree-optimization/78384
31651 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31653 2017-01-23 Richard Biener <rguenther@suse.de>
31655 PR tree-optimization/79186
31656 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31657 both incoming edges before moving an assert.
31659 2017-01-23 Martin Jambor <mjambor@suse.cz>
31661 * ipa-prop.c (load_from_param_1): Removed.
31662 (load_from_unmodified_param): Bits from load_from_param_1 put back
31664 (load_from_param): Removed.
31665 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31666 with stmt. Reverted back to use of load_from_unmodified_param.
31668 2017-01-23 Martin Jambor <mjambor@suse.cz>
31671 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31672 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31673 field a pointer to garbage collected vector, mark lattices and
31674 ipcp_orig_node with GTY((skip)).
31675 (ipa_get_param_count): Adjust to descriptors being a pointer.
31676 (ipa_get_param): Likewise.
31677 (ipa_get_type): Likewise.
31678 (ipa_get_param_move_cost): Likewise.
31679 (ipa_set_param_used): Likewise.
31680 (ipa_get_controlled_uses): Likewise.
31681 (ipa_set_controlled_uses): Likewise.
31682 (ipa_is_param_used): Likewise.
31683 (ipa_node_params_t): Move into garbage collector. New methods insert
31685 (ipa_node_params_sum): Annotate wth GTY(()).
31686 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31688 (ipa_load_from_parm_agg): Adjust declaration.
31689 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31690 * ipa-profile.c (ipa_profile): Likewise.
31691 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31692 (ipa_populate_param_decls): Make descriptors parameter garbage
31694 (ipa_dump_param): Adjust to descriptors being a pointer.
31695 (ipa_alloc_node_params): Likewise.
31696 (ipa_initialize_node_params): Likewise.
31697 (load_from_param_1): Make descriptors parameter garbage collected.
31698 (load_from_unmodified_param): Likewise.
31699 (load_from_param): Likewise.
31700 (ipa_load_from_parm_agg): Likewise.
31701 (ipa_node_params::~ipa_node_params): Removed.
31702 (ipa_free_all_node_params): Remove call to delete operator.
31703 (ipa_node_params_t::insert): New.
31704 (ipa_node_params_t::remove): Likewise.
31705 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31706 copy known_csts and known_contexts vectors.
31707 (ipa_read_node_info): Adjust to descriptors being a pointer.
31708 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31710 (ipcp_transform_function): Make descriptors variable garbage
31713 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31715 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31716 * config/i386/avx512dqintrin.h: Ditto.
31717 * config/i386/avx512fintrin.h: Ditto.
31718 * config/i386/i386.c: Handle new builtins.
31719 * config/i386/i386-builtin.def: Add new builtins.
31720 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31721 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31723 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31724 Martin Liska <mliska@suse.cz>
31726 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31727 * asan.c (asan_expand_poison_ifn): Support stores and use
31728 appropriate ASAN report function.
31729 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31730 * internal-fn.def (ASAN_POISON_USE): Declare.
31731 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31732 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31733 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31734 ASAN_POISON calls w/o LHS.
31735 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31736 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31737 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31738 * gimplify.c (asan_poison_variables): Add attribute
31739 use_after_scope_memory to variables that really needs to live
31741 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31742 having the attribute.
31744 2017-01-23 Martin Liska <mliska@suse.cz>
31746 * asan.c (create_asan_shadow_var): New function.
31747 (asan_expand_poison_ifn): Likewise.
31748 * asan.h (asan_expand_poison_ifn): New declaration.
31749 * internal-fn.c (expand_ASAN_POISON): Likewise.
31750 * internal-fn.def (ASAN_POISON): New builtin.
31751 * sanopt.c (pass_sanopt::execute): Expand
31752 asan_expand_poison_ifn.
31753 * tree-inline.c (copy_decl_for_dup_finish): Make function
31755 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31756 * tree-ssa.c (is_asan_mark_p): New function.
31757 (execute_update_addresses_taken): Rewrite local variables
31758 (identified just by use-after-scope as addressable) into SSA.
31760 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31762 * doc/install.texi (Specific): opensource.apple.com uses https
31763 now. Remove trailing slash.
31765 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31767 * README.Portability: Remove note on an Irix compatibility issue.
31769 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31771 * gcov.c (INCLUDE_ALGORITHM): Define.
31772 (INCLUDE_VECTOR): Define.
31773 No longer include <vector> and <algorithm> directly.
31775 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31777 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31779 * doc/invoke.texi (Code Gen Options): Ditto.
31781 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31784 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31786 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31788 rtl-optimization/79125
31789 * cprop.c (local_cprop_pass): Handle cases where we make an
31790 unconditional trap.
31792 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31796 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31797 read from, for big endian.
31799 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31801 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31802 register pauth builtins for LP64 only.
31804 2017-01-20 Marek Polacek <polacek@redhat.com>
31807 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31810 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31812 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31814 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31815 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31816 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31818 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31821 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31822 in big-endian mode when they are not a single duplicated value.
31824 2017-01-20 Richard Biener <rguenther@suse.de>
31826 * BASE-VER: Bump to 7.0.1.
31828 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31830 * omp-low.c (omplow_simd_context): New struct. Use it...
31831 (lower_rec_simd_input_clauses): ...here and...
31832 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31833 references to idx, lane, max_vf, is_simt.
31835 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31837 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31840 2017-01-20 Martin Jambor <mjambor@suse.cz>
31842 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31843 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31845 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31846 (GTFILES): Rename hsa.c to hsa-common.c.
31847 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31848 * hsa-dump.c: Likewise.
31849 * hsa-gen.c: Likewise.
31850 * hsa-regalloc.c: Likewise.
31851 * ipa-hsa.c: Likewise.
31852 * omp-expand.c: Likewise.
31853 * omp-low.c: Likewise.
31854 * toplev.c: Likewise.
31856 2017-01-20 Marek Polacek <polacek@redhat.com>
31859 * doc/invoke.texi: Document -Wduplicated-branches.
31860 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31861 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31862 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31863 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31864 return 0 only when not OEP_LEXICOGRAPHIC.
31865 (fold_build_cleanup_point_expr): Use the expression
31866 location when building CLEANUP_POINT_EXPR.
31867 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31868 * tree.c (add_expr): Handle error_mark_node.
31870 2017-01-20 Martin Liska <mliska@suse.cz>
31873 * tree-profile.c (init_ic_make_global_vars): Do not call
31875 (gimple_init_gcov_profiler): Likewise.
31877 2017-01-20 Martin Liska <mliska@suse.cz>
31880 * cgraph.h (maybe_create_reference): Remove argument and
31882 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31884 * ipa-cp.c (create_specialized_node): Likewise.
31885 * symtab.c (symtab_node::maybe_create_reference): Handle
31886 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31888 2017-01-20 Martin Liska <mliska@suse.cz>
31890 * read-rtl-function.c (function_reader::create_function): Use
31891 build_decl instread of build_decl_stat.
31893 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31895 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31896 * config/i386/avx512dqintrin.h: Ditto.
31897 * config/i386/avx512fintrin.h: Ditto.
31898 * config/i386/i386-builtin-types.def: Add new types.
31899 * config/i386/i386.c: Handle new types.
31900 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31901 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31902 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31903 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31904 (__builtin_ia32_kshiftridi): New.
31905 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31907 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31911 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31912 define to rs6000_init_stack_protect_guard.
31913 (rs6000_init_stack_protect_guard): New function.
31915 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31916 Yunqiang Su <yunqiang.su@imgtec.com>
31918 * config.gcc (supported_defaults): Add madd4.
31919 (with_madd4): Add validation.
31920 (all_defaults): Add madd4.
31921 * config/mips/mips.opt (mmadd4): New option.
31922 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31924 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31926 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31927 (ISA_HAS_FUSED_MADD4): Likewise.
31928 * doc/invoke.texi (-mmadd4): Document the new option.
31929 * doc/install.texi (--with-madd4): Document the new option.
31931 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31933 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31934 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31935 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31936 (aarch64_init_pauth_hint_builtins): New.
31937 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31938 (aarch64_expand_builtin): Expand new builtins.
31940 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31942 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31943 * combine-stack-adj.c (no_unhandled_cfa): Handle
31944 REG_CFA_TOGGLE_RA_MANGLE.
31945 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31946 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31947 info for return address signing.
31948 (aarch64_expand_epilogue): Likewise.
31950 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31952 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31953 * config/aarch64/aarch64-protos.h
31954 (aarch64_return_address_signing_enabled): New declaration.
31955 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31957 (aarch64_expand_prologue): Sign return address before it's pushed onto
31959 (aarch64_expand_epilogue): Authenticate return address fetched from
31961 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31962 (aarch64_attributes): New function attributes for "sign-return-address".
31963 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31964 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31965 ("*do_return"): Generate combined instructions according to key index.
31966 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31967 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31969 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31970 * config/aarch64/aarch64.opt (msign-return-address=): New.
31971 * doc/extend.texi (AArch64 Function Attributes): Documents
31972 "sign-return-address=".
31973 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31975 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31977 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31978 overall option summary.
31980 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31982 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31983 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31984 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31985 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31987 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31989 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31990 -mpower9-minmax by default for -mcpu=power9.
31991 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31992 128-bit floating point.
31994 2017-01-20 Alan Modra <amodra@gmail.com>
31996 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31997 optimizing for size.
31999 2017-01-20 Alan Modra <amodra@gmail.com>
32002 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32003 for strcmp and strncmp from corresponding builtin decl.
32005 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32007 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32008 instead of i386/rtems-64.h.
32009 * config/i386/rtems-64.h: Remove.
32011 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32015 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32017 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32019 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32021 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32022 Change int to HOST_WIDE_INT.
32023 * config/aarch64/aarch64-protos.h
32024 (aarch64_simd_gen_const_vector_dup): Likewise.
32025 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32027 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32029 * langhooks-def.h (lhd_type_for_size): New decl.
32030 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32031 * langhooks.c (lhd_type_for_size): New function, taken from
32034 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32036 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32037 define_bypass for CR latency.
32038 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32039 (power9-alu2): Add define_bypass for CR latency.
32041 (power9-mul): Update insn latency.
32042 (power9-mul-compare): Update insn latency, bypass latency and remove
32045 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32047 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32049 * config/aarch64/aarch64.md
32050 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32051 aarch64_nopcrelative_literal_loads.
32052 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32054 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32056 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32057 TARGET_LOONGSON_3A.
32058 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32060 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32063 * config.gcc (supported_defaults): Add lxc1-sxc1.
32064 (with_lxc1_sxc1): Add validation.
32065 (all_defaults): Add lxc1-sxc1.
32066 * config/mips/mips.opt (mlxc1-sxc1): New option.
32067 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32069 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32070 __mips_no_lxc1_sxc1.
32071 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32072 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32073 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32075 2017-01-19 Richard Biener <rguenther@suse.de>
32077 PR tree-optimization/72488
32078 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32079 sure to restore SSA info.
32080 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32082 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32084 PR rtl-optimization/79121
32085 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32086 of the inner type when shifting an extended value.
32088 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32091 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32092 interposable aliases.
32094 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32097 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32098 Use the evmergelohi instruction.
32099 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32100 (mov_sitf_e500_subreg8_2_be): Likewise.
32101 (mov_sitf_e500_subreg12_2_le): Likewise.
32102 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32103 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32104 (mov_sitf_e500_subreg8_2_le): Likewise.
32105 (mov_sitf_e500_subreg12_2_be): Likewise.
32107 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32109 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32110 attribute from vecsimple to vecperm.
32111 (altivec_vbpermq2): Likewise.
32113 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32116 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32118 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32119 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32120 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32121 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32122 case where N arg is SIZE_MAX.
32123 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32124 (cmpstrsi): Add pattern.
32126 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32128 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32129 __builtin_vec_revb builtins.
32130 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32131 built-in functions to support generation of the ISA 3.0 XXBR<x>
32132 vector byte reverse instructions.
32133 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32134 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32135 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32136 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32137 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32138 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32139 (P9V_BUILTIN_VEC_REVB): Likewise.
32140 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32141 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32142 (p9_xxbrq_v16qi): Likewise.
32143 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32144 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32145 (p9_xxbrh_v8hi): Likewise.
32146 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32147 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32148 vec_revb built-in functions.
32150 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32152 PR rtl-optimization/78952
32153 * config/i386/i386.md (any_extract): New code iterator.
32154 (*insvqi_2): Use any_extract for source operand.
32155 (*insvqi_3): Use any_shiftrt for source operand.
32157 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32159 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32161 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32163 2017-01-18 Matthias Klose <doko@ubuntu.com>
32165 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32167 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32169 * config/rs6000/altivec.h (vec_bperm): Change #define.
32170 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32171 (altivec_vbpermq2): New define_insn.
32172 (altivec_vbpermd): Likewise.
32173 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32174 function interface.
32175 (VBPERMD): Likewise.
32176 (VBPERM): New polymorphic function interface.
32177 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32178 Add entries for P9V_BUILTIN_VEC_VBPERM.
32179 * doc/extend.texi: Add interfaces for vec_bperm.
32181 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32183 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32184 first letter of error messages.
32185 (s390_resolve_overloaded_builtin): Likewise.
32186 * config/s390/s390.c (s390_expand_builtin): Likewise.
32187 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32188 (s390_valid_target_attribute_inner_p): Likewise.
32189 * config/s390/s390.md ("tabort"): Likewise.
32191 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32193 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32194 (ISA_AVOID_DIV_HILO): New macro.
32195 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32196 (ISA_HAS_DDIV): Likewise.
32198 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32200 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32202 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32204 * doc/invoke.texi (fabi-version): Spelling fix.
32206 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32209 * doc/invoke.texi (fabi-version): Mention mangling fix for
32212 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32215 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32217 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32220 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32221 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32223 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32225 (stack_protect_test): Ditto.
32226 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32227 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32229 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32230 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32231 -mstack-protector-guard-offset=.
32232 (RS/6000 and PowerPC Options): Ditto.
32234 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32236 * config/i386/i386.h (MASK_CLASS_P): New define.
32237 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32238 there are no registers from different register sets also when
32239 mask registers are used. Update function comment.
32240 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32241 to (*k/*r) and (*k/*km) alternatives.
32243 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32245 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32246 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32247 (EH_RETURN_HANDLER_RTX): New define.
32248 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32249 Force frame pointer in EH return functions.
32250 (aarch64_expand_epilogue): Add barrier for eh_return.
32251 (aarch64_final_eh_return_addr): Remove.
32252 (aarch64_eh_return_handler_rtx): New function.
32253 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32255 (aarch64_eh_return_handler_rtx): New prototype.
32257 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32259 * config/rs6000/altivec.h (vec_rlmi): New #define.
32260 (vec_vrlnm): Likewise.
32261 (vec_rlnm): Likewise.
32262 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32263 (UNSPEC_VRLNM): Likewise.
32264 (VIlong): New mode iterator.
32265 (altivec_vrl<VI_char>mi): New define_insn.
32266 (altivec_vrl<VI_char>nm): Likewise.
32267 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32269 (VRLDNM): Likewise.
32270 (RLNM): New polymorphic function entry.
32271 (VRLWMI): New monomorphic function entry.
32272 (VRLDMI): Likewise.
32273 (RLMI): New polymorphic function entry.
32274 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32275 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32276 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32279 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32282 * dwarf2out.c (field_byte_offset): Restore the
32283 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32284 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32285 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32288 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32291 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32293 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32296 * dwarf2out.c (add_data_member_location_attribute): For constant
32297 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32298 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32299 DW_AT_byte_size attributes.
32301 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32303 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32304 after forcing to constant memory when the code model is medium.
32306 2017-01-17 Julia Koval <julia.koval@intel.com>
32309 * config/i386/avx512fintrin.h
32310 (_mm512_i32gather_ps): Change __addr type to void const*.
32311 (_mm512_mask_i32gather_ps): Ditto.
32312 (_mm512_i32gather_pd): Ditto.
32313 (_mm512_mask_i32gather_pd): Ditto.
32314 (_mm512_i64gather_ps): Ditto.
32315 (_mm512_mask_i64gather_ps): Ditto.
32316 (_mm512_i64gather_pd): Ditto.
32317 (_mm512_mask_i64gather_pd): Ditto.
32318 (_mm512_i32gather_epi32): Ditto.
32319 (_mm512_mask_i32gather_epi32): Ditto.
32320 (_mm512_i32gather_epi64): Ditto.
32321 (_mm512_mask_i32gather_epi64): Ditto.
32322 (_mm512_i64gather_epi32): Ditto.
32323 (_mm512_mask_i64gather_epi32): Ditto.
32324 (_mm512_i64gather_epi64): Ditto.
32325 (_mm512_mask_i64gather_epi64): Ditto.
32326 (_mm512_i32scatter_ps): Change __addr type to void*.
32327 (_mm512_mask_i32scatter_ps): Ditto.
32328 (_mm512_i32scatter_pd): Ditto.
32329 (_mm512_mask_i32scatter_pd): Ditto.
32330 (_mm512_i64scatter_ps): Ditto.
32331 (_mm512_mask_i64scatter_ps): Ditto.
32332 (_mm512_i64scatter_pd): Ditto.
32333 (_mm512_mask_i64scatter_pd): Ditto.
32334 (_mm512_i32scatter_epi32): Ditto.
32335 (_mm512_mask_i32scatter_epi32): Ditto.
32336 (_mm512_i32scatter_epi64): Ditto.
32337 (_mm512_mask_i32scatter_epi64): Ditto.
32338 (_mm512_i64scatter_epi32): Ditto.
32339 (_mm512_mask_i64scatter_epi32): Ditto.
32340 (_mm512_i64scatter_epi64): Ditto.
32341 (_mm512_mask_i64scatter_epi64): Ditto.
32342 * config/i386/avx512pfintrin.h
32343 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32344 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32345 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32346 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32347 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32348 (_mm512_prefetch_i32scatter_ps): Ditto.
32349 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32350 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32351 (_mm512_prefetch_i64scatter_pd): Ditto.
32352 (_mm512_prefetch_i64scatter_ps): Ditto.
32353 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32354 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32355 * config/i386/avx512vlintrin.h
32356 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32357 (_mm_mmask_i32gather_ps): Ditto.
32358 (_mm256_mmask_i32gather_pd): Ditto.
32359 (_mm_mmask_i32gather_pd): Ditto.
32360 (_mm256_mmask_i64gather_ps): Ditto.
32361 (_mm_mmask_i64gather_ps): Ditto.
32362 (_mm256_mmask_i64gather_pd): Ditto.
32363 (_mm_mmask_i64gather_pd): Ditto.
32364 (_mm256_mmask_i32gather_epi32): Ditto.
32365 (_mm_mmask_i32gather_epi32): Ditto.
32366 (_mm256_mmask_i32gather_epi64): Ditto.
32367 (_mm_mmask_i32gather_epi64): Ditto.
32368 (_mm256_mmask_i64gather_epi32): Ditto.
32369 (_mm_mmask_i64gather_epi32): Ditto.
32370 (_mm256_mmask_i64gather_epi64): Ditto.
32371 (_mm_mmask_i64gather_epi64): Ditto.
32372 (_mm256_i32scatter_ps): Change __addr type to void*.
32373 (_mm256_mask_i32scatter_ps): Ditto.
32374 (_mm_i32scatter_ps): Ditto.
32375 (_mm_mask_i32scatter_ps): Ditto.
32376 (_mm256_i32scatter_pd): Ditto.
32377 (_mm256_mask_i32scatter_pd): Ditto.
32378 (_mm_i32scatter_pd): Ditto.
32379 (_mm_mask_i32scatter_pd): Ditto.
32380 (_mm256_i64scatter_ps): Ditto.
32381 (_mm256_mask_i64scatter_ps): Ditto.
32382 (_mm_i64scatter_ps): Ditto.
32383 (_mm_mask_i64scatter_ps): Ditto.
32384 (_mm256_i64scatter_pd): Ditto.
32385 (_mm256_mask_i64scatter_pd): Ditto.
32386 (_mm_i64scatter_pd): Ditto.
32387 (_mm_mask_i64scatter_pd): Ditto.
32388 (_mm256_i32scatter_epi32): Ditto.
32389 (_mm256_mask_i32scatter_epi32): Ditto.
32390 (_mm_i32scatter_epi32): Ditto.
32391 (_mm_mask_i32scatter_epi32): Ditto.
32392 (_mm256_i32scatter_epi64): Ditto.
32393 (_mm256_mask_i32scatter_epi64): Ditto.
32394 (_mm_i32scatter_epi64): Ditto.
32395 (_mm_mask_i32scatter_epi64): Ditto.
32396 (_mm256_i64scatter_epi32): Ditto.
32397 (_mm256_mask_i64scatter_epi32): Ditto.
32398 (_mm_i64scatter_epi32): Ditto.
32399 (_mm_mask_i64scatter_epi32): Ditto.
32400 (_mm256_i64scatter_epi64): Ditto.
32401 (_mm256_mask_i64scatter_epi64): Ditto.
32402 (_mm_i64scatter_epi64): Ditto.
32403 (_mm_mask_i64scatter_epi64): Ditto.
32404 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32405 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32406 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32407 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32408 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32409 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32410 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32411 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32412 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32413 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32414 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32415 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32416 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32417 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32418 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32419 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32420 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32421 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32422 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32423 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32424 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32425 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32426 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32427 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32428 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32429 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32430 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32431 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32432 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32433 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32434 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32435 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32436 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32437 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32438 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32439 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32440 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32441 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32442 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32443 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32444 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32445 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32446 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32447 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32448 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32449 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32450 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32451 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32452 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32453 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32454 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32455 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32456 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32457 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32458 definitions accordingly.
32460 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32461 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32464 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32467 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32470 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32471 hard regs for inner regno.
32473 2017-01-17 Martin Liska <mliska@suse.cz>
32476 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32477 assumption and add comment.
32479 2017-01-17 Nathan Sidwell <nathan@acm.org>
32481 * ipa-visibility.c (localize_node): New function, broken out of ...
32482 (function_and_variable_visibility): ... here. Call it.
32484 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32486 PR middle-end/77445
32487 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32488 correctly set frequency of oudgoing edge.
32489 (duplicate_thread_path): Fix profile updating.
32491 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32494 * configure.ac: Add GCC_BASE_VER.
32495 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32496 version from BASE-VER file.
32497 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32498 (gcc.o): Depend on $(BASEVER).
32499 * common.opt (dumpfullversion): New option.
32500 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32501 * doc/invoke.texi: Document -dumpfullversion.
32502 * doc/install.texi: Document --with-gcc-major-version-only.
32503 * configure: Regenerated.
32505 2017-01-17 Richard Biener <rguenther@suse.de>
32507 PR tree-optimization/71433
32508 * tree-vrp.c (register_new_assert_for): Merge same asserts
32509 on all incoming edges.
32510 (process_assert_insertions_for): Handle insertions at the
32513 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32515 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32516 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32518 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32521 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32524 2017-01-17 Alan Modra <amodra@gmail.com>
32527 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32528 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32529 symbolic stack limit when pic.
32531 2017-01-16 Martin Sebor <msebor@redhat.com>
32533 PR tree-optimization/78608
32534 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32536 2017-01-16 Jeff Law <law@redhat.com>
32539 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32540 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32541 for several include directories that may be relative to sysroot.
32542 * config/i386/x-mingw32 (gplus_includedir): Define.
32543 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32544 (native_system_includedir): Likewise.
32545 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32546 override if TARGET_SYSTEM_ROOT is defined.
32547 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32549 PR tree-optimization/79090
32550 PR tree-optimization/33562
32551 PR tree-optimization/61912
32552 PR tree-optimization/77485
32553 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32554 and computed trims into the dump file.
32556 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32558 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32560 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32563 * gimplify.c (gimplify_init_constructor): If want_value and
32564 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32568 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32569 sequence. Formatting fixes.
32570 (doloop_optimize): Formatting fixes.
32573 * gcc.c (debug_level_greater_than_spec_func): New function.
32574 (static_spec_functions): Add debug-level-gt spec function.
32575 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32577 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32578 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32579 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32580 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32581 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32582 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32584 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32586 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32587 QImode fixups to general and mask registers only.
32589 2017-01-16 Carl Love <cel@us.ibm.com>
32591 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32592 for built-in functions
32593 vector signed char vec_nabs (vector signed char)
32594 vector signed short vec_nabs (vector signed short)
32595 vector signed int vec_nabs (vector signed int)
32596 vector signed long long vec_nabs (vector signed long long)
32597 vector float vec_nabs (vector float)
32598 vector double vec_nabs (vector double)
32599 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32601 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32602 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32603 * doc/extend.texi: Update the documentation file for the new built-in
32606 2017-01-16 Martin Sebor <msebor@redhat.com>
32608 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32611 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32613 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32614 UNSPEC_VSX__XXSPLTD to require special splat handling.
32616 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32619 * system.h: Poison strndup.
32621 2017-01-16 Alan Modra <amodra@gmail.com>
32624 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32627 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32629 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32631 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32633 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32634 call recog here. Assert that INSN_CODE (insn) is non-negative.
32636 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32639 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32641 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32642 in the currently scheduled RTL fragment.
32644 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32646 PR rtl-optimization/78751
32647 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32650 2017-01-14 Jeff Law <law@redhat.com>
32652 PR tree-optimization/79090
32653 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32654 variable length stores.
32655 (compute_trims): Delete dead assignment to *trim_tail.
32656 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32659 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32661 PR rtl-optimization/78626
32662 PR rtl-optimization/78727
32663 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32664 of a block, and split such blocks after everything else is finished.
32666 2017-01-14 Alan Modra <amodra@gmail.com>
32669 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32670 target legitimate_combined_insn.
32671 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32672 (rs6000_legitimate_combined_insn): New function.
32673 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32675 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32676 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32677 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32679 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32681 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32683 2017-01-13 Jeff Law <law@redhat.com>
32685 PR tree-optimization/33562
32686 PR tree-optimization/61912
32687 PR tree-optimization/77485
32688 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32690 (delete_dead_assignment): Likewise.
32691 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32692 statement to delete_dead_call and delete_dead_assignment.
32694 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32697 * substring-locations.c (format_warning_va): Strengthen case 1 so
32698 that both endpoints of the substring must be within the format
32699 range for just the substring to be printed.
32701 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32703 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32704 * config/i386/i386.c (ix86_target_string): Add missing options
32705 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32706 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32707 flags_other and ix86_target_other to flags2_other. Display unknown
32709 (ix86_valid_target_attribute_inner_p): Add missing options and
32710 reorder options by implied ISAs, as in ix86_target_string.
32712 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32714 * hash-table.h (hash_table::too_empty_p): New function.
32715 (hash_table::expand): Use it.
32716 (hash_table::traverse): Likewise.
32717 (hash_table::empty_slot): Use sizeof (value_type) instead of
32718 sizeof (PTR) to convert bytes to elements. Shrink the table
32719 if the current size is excessive for the current number of
32722 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32724 * ira-costs.c (record_reg_classes): Break from the inner loop
32725 early once alt_fail is known to be true. Update outer loop
32726 handling accordingly.
32728 2017-01-13 Jeff Law <law@redhat.com>
32730 * tree-ssa-dse.c (decrement_count): New function.
32731 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32732 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32733 when we know the partially dead statement is a mem* function.
32735 PR tree-optimization/61912
32736 PR tree-optimization/77485
32737 * tree-ssa-dse.c: Include expr.h.
32738 (maybe_trim_constructor_store): New function.
32739 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32741 PR tree-optimization/33562
32742 PR tree-optimization/61912
32743 PR tree-optimization/77485
32744 * doc/invoke.texi: Document new dse-max-object-size param.
32745 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32746 * tree-ssa-dse.c: Include params.h.
32747 (dse_store_status): New enum.
32748 (initialize_ao_ref_for_dse): New, partially extracted from
32750 (valid_ao_ref_for_dse, normalize_ref): New.
32751 (setup_live_bytes_from_ref, compute_trims): Likewise.
32752 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32753 (maybe_trim_partially_dead_store): Likewise.
32754 (maybe_trim_complex_store): Likewise.
32755 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32756 Track what bytes live from the original store. Return tri-state
32757 for dead, partially dead or live.
32758 (dse_dom_walker): Add constructor, destructor and new private members.
32759 (delete_dead_call, delete_dead_assignment): New extracted from
32761 (dse_optimize_stmt): Make a member of dse_dom_walker.
32762 Use initialize_ao_ref_for_dse.
32764 PR tree-optimization/33562
32765 PR tree-optimization/61912
32766 PR tree-optimization/77485
32767 * sbitmap.h (bitmap_count_bits): Prototype.
32768 (bitmap_clear_range, bitmap_set_range): Likewise.
32769 * sbitmap.c (bitmap_clear_range): New function.
32770 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32772 2017-01-13 Martin Liska <mliska@suse.cz>
32775 * function.c (set_cfun): Add new argument force.
32776 * function.h (set_cfun): Likewise.
32777 * ipa-inline-transform.c (inline_call): Use the function when
32778 strict alising from is dropped for function we inline to.
32780 2017-01-13 Richard Biener <rguenther@suse.de>
32782 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32783 for dumping GIMPLE INTEGER_CSTs.
32785 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32787 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32788 to 201112L since C++17.
32790 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32793 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32794 if -fsanitize=kernel-address is present.
32796 2017-01-13 Richard Biener <rguenther@suse.de>
32798 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32799 as _Literal ( type ) number in case usual suffixes do not
32800 preserve all information.
32802 2017-01-13 Richard Biener <rguenther@suse.de>
32804 PR tree-optimization/77283
32805 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32806 and ssa-iterators.h.
32807 (is_feasible_trace): Implement a cost model based on joiner
32810 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32813 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32814 char or short to __float128/_Float128 directly.
32816 2017-01-12 Martin Sebor <msebor@redhat.com>
32818 to -Wformat-overflow.
32819 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32820 (min_bytes_remaining): Same.
32821 (get_string_length): Same.
32822 (format_string): Same.
32823 (format_directive): Same.
32825 (pass_sprintf_length::handle_gimple_call): Same.
32827 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32829 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32830 info.nowrite calls with no lhs that can't throw. Return bool
32831 whether gsi_remove has been called or not.
32832 (pass_sprintf_length::handle_gimple_call): Return bool whether
32833 try_substitute_return_value called gsi_remove. Formatting fix.
32834 (pass_sprintf_length::execute): Don't use gsi_remove if
32835 handle_gimple_call returned true.
32838 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32839 be removed due to side-effects, don't remove following barrier nor
32840 turn the successor edge into fallthru edge.
32842 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32845 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32846 element-reversing loads and stores as not swappable.
32848 2017-01-12 Nathan Sidwell <nathan@acm.org>
32849 Nicolai Stange <nicstange@gmail.com>
32851 * combine.c (try_combine): Don't ignore result of overlap checking
32852 loop. Combine overlap & asm check into single loop.
32854 2017-01-12 Richard Biener <rguenther@suse.de>
32856 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32857 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32859 2017-01-12 Richard Biener <rguenther@suse.de>
32861 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32862 and TS_TARGET_OPTION directly derive from TS_BASE.
32863 * tree-core.h (tree_optimization_option): Derive from tree_base.
32864 (tree_target_option): Likewise.
32866 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32868 * config/i386/i386.c (memory_address_length): Increase len
32869 only when rip_relative_addr_p returns false.
32871 2017-01-11 Julia Koval <julia.koval@intel.com>
32873 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32874 (OPTION_MASK_ISA_SGX_SET): New.
32875 (ix86_handle_option): Handle OPT_msgx.
32876 * config.gcc: Added sgxintrin.h.
32877 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32878 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32879 * config/i386/i386.c (ix86_target_string): Add -msgx.
32881 (ix86_option_override_internal): Handle new options.
32882 (ix86_valid_target_attribute_inner_p): Add sgx.
32883 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32884 * config/i386/i386.opt: Add msgx.
32885 * config/i386/sgxintrin.h: New file.
32886 * config/i386/x86intrin.h: Add sgxintrin.h.
32888 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32891 * fold-const.c (maybe_nonzero_address): Return 1 for function
32893 (tree_single_nonzero_warnv_p): Don't handle function local objects
32897 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32900 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32903 * opts.c: Include "spellcheck.h"
32904 (struct string_fragment): New struct.
32905 (struct edit_distance_traits<const string_fragment &>): New
32907 (get_closest_sanitizer_option): New function.
32908 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32910 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32912 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32914 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32915 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32916 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32917 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32918 for initial die_offset if dwarf_split_debug_info.
32919 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32920 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32922 (output_skeleton_debug_sections): Formatting fix. Use
32923 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32924 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32926 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32928 * config/arm/cortex-a53.md: Add bypasses for
32929 cortex_a53_r2f_cvt.
32930 (cortex_a53_r2f): Only use for transfers.
32931 (cortex_a53_f2r): Likewise.
32932 (cortex_a53_r2f_cvt): Add reservation for conversions.
32933 (cortex_a53_f2r_cvt): Likewise.
32935 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32937 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32938 to all inlined functions, change static to extern.
32940 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32943 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32945 (arm_assemble_integer): Likewise.
32947 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32949 * config.gcc: Use new awk script to check CPU, FPU and architecture
32950 parameters for --with-... options.
32951 * config/arm/parsecpu.awk: New file
32952 * config/arm/arm-cpus.in: New file.
32953 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32955 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32957 * config/arm/t-arm: Update dependency rules.
32958 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32959 of processing .def files.
32960 * config/arm/genopt.sh: Deleted.
32961 * config/arm/gentune.sh: Deleted.
32962 * config/arm/arm-cores.def: Deleted.
32963 * config/arm/arm-arches.def: Deleted.
32964 * config/arm/arm-fpus.def: Deleted.
32965 * config/arm/arm-tune.md: Regenerated.
32966 * config/arm/arm-tables.opt: Regenerated.
32967 * config/arm/arm-cpu.h: New generated file.
32968 * config/arm/arm-cpu-data.h: New generated file.
32969 * config/arm/arm-cpu-cdata.h: New generated file.
32971 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32974 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32976 (input_varpool_node): Unpack dynamically_initialized bit.
32978 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32980 PR rtl-optimization/79032
32981 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32982 the alignment of the adjusted memory reference against that of MODE,
32983 instead of the alignment of the original memory reference.
32985 2017-01-11 Martin Jambor <mjambor@suse.cz>
32987 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32989 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32990 decorated functions.
32992 2017-01-11 Richard Biener <rguenther@suse.de>
32994 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32995 set range/nonnull info for PHI results. Do not set it on
32996 stmts marked for removal.
32998 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33000 * expr.c (store_field): In the bitfield case, fetch the return value
33001 from the registers before applying a single big-endian adjustment.
33002 Always do a final load for a BLKmode value not larger than a word.
33004 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33007 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33008 that we correctly handle column numbers greater than
33009 LINE_MAP_MAX_COLUMN_NUMBER.
33011 2017-01-10 Martin Sebor <msebor@redhat.com>
33013 PR middle-end/78245
33014 * gimple-ssa-sprintf.c (get_destination_size): Call
33015 {init,fini}object_sizes.
33016 * tree-object-size.c (addr_object_size): Adjust.
33017 (pass_through_call): Adjust.
33018 (pass_object_sizes::execute): Adjust.
33019 * tree-object-size.h (fini_object_sizes): Declare.
33021 2017-01-10 Martin Sebor <msebor@redhat.com>
33023 PR tree-optimization/78775
33024 * builtins.c (get_size_range): Move...
33025 * calls.c: ...to here.
33026 (alloc_max_size): Accept zero argument.
33027 (operand_signed_p): Remove.
33028 (maybe_warn_alloc_args_overflow): Call get_size_range.
33029 * calls.h (get_size_range): Declare.
33031 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33033 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33034 from TI's devices.csv file as of September 2016.
33035 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33037 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33039 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33040 * doc/invoke.texi: Likewise.
33041 * doc/md.texi: Likewise.
33042 * doc/objc.texi: Likewise.
33044 2017-01-10 Joshua Conner <joshconner@google.com>
33046 * config/arm/fuchsia-elf.h: New file.
33047 * config/fuchsia.h: New file.
33048 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33049 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33051 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33053 2016-01-10 Richard Biener <rguenther@suse.de>
33055 PR tree-optimization/79034
33056 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33057 Propagate out degenerate PHIs in the joiner.
33059 2017-01-10 Martin Liska <mliska@suse.cz>
33061 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33062 (sort_congruence_classes_by_decl_uid): Likewise.
33063 (sort_congruence_class_groups_by_decl_uid): Likewise.
33064 (sem_item_optimizer::merge_classes): Sort class, groups in these
33065 classes and members in the groups by DECL_UID of declarations.
33066 This would make merge operations stable.
33068 2017-01-10 Martin Liska <mliska@suse.cz>
33070 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33071 usage of m_classes_vec.
33072 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33073 (sem_item_optimizer::get_group_by_hash): Likewise.
33074 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33075 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33076 (sem_item_optimizer::verify_classes): Likewise.
33077 (sem_item_optimizer::process_cong_reduction): Likewise.
33078 (sem_item_optimizer::dump_cong_classes): Likewise.
33079 (sem_item_optimizer::merge_classes): Likewise.
33080 * ipa-icf.h (congruence_class_hash): Rename from
33081 congruence_class_group_hash. Remove declaration of m_classes_vec.
33083 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33085 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33086 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33087 * config.gcc: Add avx512vpopcntdqintrin.h.
33088 * config/i386/avx512vpopcntdqintrin.h: New.
33089 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33090 * config/i386/i386-builtin-types.def: Add new types.
33091 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33092 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33093 __builtin_ia32_vpopcountq_v8di_mask): New.
33094 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33095 __AVX512VPOPCNTDQ__.
33096 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33097 (PTA_AVX512VPOPCNTDQ): Define.
33098 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33099 TARGET_AVX512VPOPCNTDQ_P): Define.
33100 * config/i386/i386.opt: Add mavx512vpopcntdq.
33101 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33102 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33104 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33106 PR middle-end/77484
33107 * predict.def (PRED_CALL): Set to 67.
33109 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33111 * expr.c (store_field): In the bitfield case, if the value comes from
33112 a function call and is of an aggregate type returned in registers, do
33113 not modify the field mode; extract the value in all cases if the mode
33114 is BLKmode and the size is not larger than a word.
33116 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33119 * config/i386/cpuid.h: Fix undefined behavior.
33121 2017-01-04 Jeff Law <law@redhat.com>
33123 PR tree-optimization/79007
33124 PR tree-optimization/67955
33125 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33126 conservative for pt.null when flag_non_call_exceptions is on.
33128 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33130 PR translation/79019
33131 PR translation/79020
33132 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33133 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33134 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33136 * config/avr/avr.opt (maccumulate-args): Likewise.
33137 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33138 * common.opt (freport-bug): Likewise.
33139 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33140 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33141 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33142 translatable string.
33143 * config/i386/i386.c (function_value_32): Likewise.
33144 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33145 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33147 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33148 * common/config/msp430/msp430-common.c (msp430_handle_option):
33150 * symtab.c (symtab_node::verify_base): Likewise.
33151 * opts.c (set_debug_level): Likewise.
33152 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33153 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33154 missing whitespace to translatable strings.
33155 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33156 * config/sh/superh.h: Likewise.
33157 * config/i386/xopintrin.h: Likewise.
33158 * config/i386/znver1.md: Likewise.
33159 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33160 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33161 * double-int.h (struct double_int): Likewise.
33162 * double-int.c (div_and_round_double): Likewise.
33163 * wide-int.cc: Likewise.
33164 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33165 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33166 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33167 (crossjumps_occurred): ... this.
33168 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33171 PR tree-optimization/78899
33172 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33173 returning bool return struct loop *, NULL for failure and the new
33175 (versionable_outer_loop_p): Don't version outer loop if it has
33176 dont_vectorized bit set.
33177 (tree_if_conversion): When versioning outer loop, ensure
33178 tree_if_conversion is performed also on the inner loop of the
33179 non-vectorizable outer loop copy.
33180 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33181 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33182 prevent vectorization of it.
33183 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33184 the outer loop vectorization of the non-scalar version is attempted
33185 before vectorization of the inner loop in scalar version. If
33186 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33187 vectorization of its inner loop.
33188 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33189 has 2 inner loops, rename also on edges from bb whose single pred
33190 is outer_loop->header. Fix typo in function comment.
33192 2017-01-09 Martin Sebor <msebor@redhat.com>
33195 * asan.c (asan_emit_stack_protection): Increase local buffer size
33196 to avoid snprintf truncation warning.
33198 2017-01-09 Andrew Pinski <apinski@cavium.com>
33200 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33201 to reference thunderx2t99 for the tuning structure
33202 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33204 (thunderx2t99_extra_costs): This.
33205 * config/aarch64/aarch64-tune.md: Regenerate.
33206 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33207 (vulcan_addrcost_table): This.
33208 (vulcan_regmove_cost): Rename to ...
33209 (thunderx2t99_regmove_cost): This.
33210 (vulcan_vector_cost): Rename to ...
33211 (thunderx2t99_vector_cost): this.
33212 (vulcan_branch_cost): Rename to ...
33213 (thunderx2t99_branch_cost): This.
33214 (vulcan_tunings): Rename to ...
33215 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33216 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33218 2017-01-09 Martin Jambor <mjambor@suse.cz>
33222 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33223 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33224 (propagate_vr_accross_jump_function): Use the above function for all
33225 value range computations for pass-through jump functions and type
33226 converasion from explicit value range values.
33227 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33228 parameters from TYPE_ARG_TYPES.
33229 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33230 (ipa_write_node_info): Stream type of the actual argument.
33231 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33233 2017-01-09 Martin Liska <mliska@suse.cz>
33236 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33237 (lookup_compiler): Do not show error message with have_E.
33239 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33241 PR tree-optimization/78938
33242 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33243 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33244 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33245 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33248 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33250 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33253 2017-01-09 Richard Biener <rguenther@suse.de>
33255 PR tree-optimization/78997
33256 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33257 name condition properly.
33259 2017-01-09 Richard Biener <rguenther@suse.de>
33262 * dwarf2out.c (is_cxx): New overload with context.
33263 (is_naming_typedef_decl): Use it.
33265 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33267 * invoke.texi (Option Summary): Correct spacing in option lists
33268 and add line breaks to fix over-long lines.
33270 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33272 PR middle-end/17660
33274 * extend.texi (Common Variable Attributes): Add xref to GCC
33275 Internals manual to explain mode attribute keywords.
33277 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33280 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33281 and Preprocessor Options.
33282 (Options for Linking): Document -pthread here....
33283 (RS/6000 and PowerPC Options): ...not here.
33284 (Solaris 2 Options): ...or here.
33285 * doc/cppopts.texi: Document -pthread.
33287 2017-01-08 Martin Sebor <msebor@redhat.com>
33289 PR middle-end/77708
33290 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33291 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33292 New member functions.
33293 (format_directive): Used them.
33295 (pass_sprintf_length::handle_gimple_call): Same.
33296 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33297 to avoid truncation for any argument.
33298 (extract_affine_mul): Same.
33299 * tree.c (get_file_function_name): Same.
33301 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33303 PR middle-end/77484
33304 * predict.def (PRED_INDIR_CALL): Set to 86.
33306 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33308 PR preprocessor/54124
33309 * doc/cppopts.texi: Reformat -d subtable to list the full name
33310 of the options. Add cross-reference to the docs for the general
33311 compiler -d options.
33312 * doc/invoke.texi (Developer Options): Add cross-reference to the
33313 preprocessor-specific -d option documentation.
33315 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33317 PR preprocessor/13498
33318 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33319 redudant material, and reflect new command-line options.
33320 (System Headers): Likewise.
33322 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33324 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33325 -isystem, and -idirafter. Copy-edit.
33326 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33327 default for -ftrack-macro-expansion. Delete obsolete and
33328 badly-formatted implementation details about -fdebug-cpp output.
33329 * doc/cppwarnopts.texi: Copy-edit.
33331 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33334 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33335 that the transition from a max line width >= 1<<10 to narrower
33336 lines works correctly.
33338 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33340 * doc/options.texi (PerFunction): New.
33341 * opt-functions.awk (switch_flags): Map both Optimization and
33342 PerFunction to CL_OPTIMIZATION.
33343 * opth-gen.awk: Test for PerFunction flag along with
33345 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33346 it only when the latter is present. Skip those that don't in
33347 the hash function generator.
33348 * common.opt (fvar-tracking): Mark as PerFunction instead of
33350 (fvar-tracking-assignments): Likewise.
33351 (fvar-tracking-assignments-toggle): Likewise.
33352 (fvar-tracking-uninit): Likewise.
33354 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33356 PR translation/79018
33357 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33360 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33363 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33364 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33365 TARGET_LONG_JUMP_TABLE_OFFSETS.
33366 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33367 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33368 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33369 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33370 * config/m68k/m68k.md (tablejump expander): Likewise.
33371 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33372 TARGET_LONG_JUMP_TABLE_OFFSETS.
33373 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33374 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33376 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33377 David Holsgrove <david.holsgrove@xilinx.com>
33379 * common/config/microblaze/microblaze-common.c
33380 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33381 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33383 * config/microblaze/microblaze.c (microblaze_must_save_register)
33384 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33386 (microblaze_eh_return): New function.
33387 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33388 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33389 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33390 * config/microblaze/microblaze.md (eh_return): New pattern.
33392 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33394 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33395 GCC_DIAGNOSTIC_STRINGIFY): Define.
33397 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33399 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33401 * config/arm/arm.md (<mcrr>): New.
33403 * config/arm/arm.c (arm_arch5te): New.
33404 (arm_option_override): Set arm_arch5te.
33405 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33407 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33408 (arm_mcrr_qualifiers): ... this. New.
33409 (MRRC_QUALIFIERS): Define to...
33410 (arm_mrrc_qualifiers): ... this. New.
33411 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33413 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33414 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33415 (MRRCI, mrrc, MRRC): New.
33416 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33417 VUNSPEC_MRRC2): New.
33419 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33421 * config/arm/arm.md (<mcr>): New.
33423 * config/arm/arm.c (arm_coproc_builtin_available): Add
33424 support for mcr, mrc, mcr2 and mrc2.
33425 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33426 (arm_mcr_qualifiers): ... this. New.
33427 (MRC_QUALIFIERS): Define to ...
33428 (arm_mrc_qualifiers): ... this. New.
33429 (MCR_QUALIFIERS): Define to ...
33430 (arm_mcr_qualifiers): ... this. New.
33431 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33433 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33434 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33435 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33436 VUNSPEC_MRC2): New.
33438 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33440 * config/arm/arm.md (*ldc): New.
33444 * config/arm/arm.c (arm_coproc_builtin_available): Add
33445 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33446 (arm_coproc_ldc_stc_legitimate_address): New.
33447 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33448 'qualifier_const_pointer'.
33449 (LDC_QUALIFIERS): Define to...
33450 (arm_ldc_qualifiers): ... this. New.
33451 (STC_QUALIFIERS): Define to...
33452 (arm_stc_qualifiers): ... this. New.
33453 * config/arm/arm-protos.h
33454 (arm_coproc_ldc_stc_legitimate_address): New.
33455 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33456 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33457 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33458 stc2, stcl, stc2l): New.
33459 * config/arm/constraints.md (Uz): New.
33460 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33461 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33462 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33463 VUNSPEC_STC2L): New.
33465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33467 * config/arm/arm.md (<cdp>): New.
33468 * config/arm/arm.c (neon_const_bounds): Rename this ...
33469 (arm_const_bounds): ... this.
33470 (arm_coproc_builtin_available): New.
33471 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33472 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33473 (CDP_QUALIFIERS): Define to...
33474 (arm_cdp_qualifiers): ... this. New.
33476 (arm_expand_builtin_args): Add case for 6 arguments.
33477 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33478 (arm_const_bounds): ... this.
33479 (arm_coproc_builtin_available): New.
33480 * config/arm/arm_acle.h (__arm_cdp): New.
33482 * config/arm/arm_acle_builtins.def (cdp): New.
33484 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33485 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33486 'arm_const_bounds'.
33487 * config/arm/types.md (coproc): New.
33488 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33489 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33490 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33491 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33493 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33495 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33496 (UBINOP_QUALIFIERS): New.
33498 (acle_builtin_data): New. Change comment.
33499 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33500 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33501 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33502 arm_acle_builtins.def.
33503 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33504 (arm_init_acle_builtins): New.
33505 (CRC32_BUILTIN): Remove.
33506 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33507 crc32cb, crc32ch and crc32cw.
33508 (arm_init_crc32_builtins): Remove.
33509 (arm_init_builtins): Use arm_init_acle_builtins rather
33510 than arm_init_crc32_builtins.
33511 (arm_expand_acle_builtin): New.
33512 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33513 * config/arm/arm_acle_builtins.def: New.
33515 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33517 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33518 (arm_builtin_datum): ... this.
33519 (arm_init_neon_builtin): Rename to ...
33520 (arm_init_builtin): ... this. Add a new parameters PREFIX
33521 and USE_SIG_IN_NAME.
33522 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33523 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33524 'arm_builtin_datum'.
33525 (arm_init_vfp_builtins): Likewise.
33526 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33527 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33528 (arm_expand_neon_args): Rename to ...
33529 (arm_expand_builtin_args): ... this. Rename builtin_arg
33530 enum values and differentiate between ARG_BUILTIN_MEMORY
33531 and ARG_BUILTIN_NEON_MEMORY.
33532 (arm_expand_neon_builtin_1): Rename to ...
33533 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33534 values, arm_expand_builtin_args and add bool parameter NEON.
33535 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33536 (arm_expand_vfp_builtin): Likewise.
33537 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33539 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33541 PR middle-end/77484
33542 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33543 * predict.c (tree_estimate_probability_bb): Reverse direction of
33544 polymorphic call predictor.
33546 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33548 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33549 (determine_pass_name_match): ...this new function and...
33550 (should_skip_pass_p): ...this new function.
33552 2017-01-06 Nathan Sidwell <nathan@acm.org>
33554 * ipa-visibility.c (function_and_variable_visibility): Reformat
33555 comments and long lines. Remove extrneous if.
33556 * symtab.c (symtab_node::make_decl_local): Fix code format.
33557 (symtab_node::set_section_for_node): Fix comment typo.
33559 2017-01-06 Martin Liska <mliska@suse.cz>
33562 * lra-constraints.c: Rename invariant to lra_invariant.
33563 * predict.c (set_even_probabilities): Initialize e to NULL.
33565 2017-01-05 Martin Sebor <msebor@redhat.com>
33567 PR tree-optimization/78910
33568 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33569 (format_integer): Correct off-by-one error in the handling
33570 of precision with negative numbers in signed conversions..
33572 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33574 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33576 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33578 PR tree-optimization/71016
33579 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33580 factor_out_conditional_conversion. Formatting fix.
33581 (factor_out_conditional_conversion): Add cond_stmt argument.
33582 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33583 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33586 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33588 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33589 read-rtl-function.o, and selftest-rtl.o.
33590 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33591 (selftest::aarch64_test_loading_full_dump): New function.
33592 (selftest::aarch64_run_selftests): New function.
33593 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33594 selftest::aarch64_run_selftests.
33595 * config/i386/i386.c
33596 (selftest::ix86_test_loading_dump_fragment_1): New function.
33597 (selftest::ix86_test_loading_call_insn): New function.
33598 (selftest::ix86_test_loading_full_dump): New function.
33599 (selftest::ix86_test_loading_unspec): New function.
33600 (selftest::ix86_run_selftests): Call the new functions.
33601 * emit-rtl.c (maybe_set_max_label_num): New function.
33602 * emit-rtl.h (maybe_set_max_label_num): New decl.
33603 * function.c (instantiate_decls): Guard call to
33604 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33605 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33607 * gensupport.c (gen_reader::gen_reader): Pass "false"
33608 for new "compact" param of rtx_reader.
33609 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33610 rather than an empty string for NULL strings.
33611 * read-md.c: Potentially include config.h rather than bconfig.h.
33612 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33613 (have_error): New global, copied from errors.c.
33614 (md_reader::read_name): Rename to...
33615 (md_reader::read_name_1): ...this, adding "out_loc" param,
33616 and converting "missing name or number" to returning false, rather
33618 (md_reader::read_name): Reimplement in terms of read_name_1.
33619 (md_reader::read_name_or_nil): New function.
33620 (md_reader::read_string): Handle "(nil)" by returning NULL.
33621 (md_reader::md_reader): Add new param "compact".
33622 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33623 (md_reader::read_file): New method.
33624 * read-md.h (md_reader::md_reader): Add new param "compact".
33625 (md_reader::read_file): New method.
33626 (md_reader::is_compact): New accessor.
33627 (md_reader::read_name): Convert return type from void to file_location.
33628 (md_reader::read_name_or_nil): New decl.
33629 (md_reader::read_name_1): New decl.
33630 (md_reader::m_compact): New field.
33631 (noop_reader::noop_reader): Pass "false" for new "compact" param
33633 (rtx_reader::rtx_reader): Add new "compact" param.
33634 (rtx_reader::read_rtx_operand): Make virtual and convert return
33635 type from void to rtx.
33636 (rtx_reader::read_until): New decl.
33637 (rtx_reader::handle_any_trailing_information): New virtual function.
33638 (rtx_reader::postprocess): New virtual function.
33639 (rtx_reader::finalize_string): New virtual function.
33640 (rtx_reader::m_in_call_function_usage): New field.
33641 (rtx_reader::m_reuse_rtx_by_id): New field.
33642 * read-rtl-function.c: New file.
33643 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33644 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33645 (selftest::verify_three_block_rtl_cfg): New decl.
33646 * read-rtl-function.h: New file.
33647 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33648 For host, include function.h, memmodel.h, and emit-rtl.h.
33649 (one_time_initialization): New function.
33650 (struct compact_insn_name): New struct.
33651 (compact_insn_names): New array.
33652 (find_code): Handle insn codes in compact dumps.
33653 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33654 (bind_subst_iter_and_attr): Likewise.
33655 (add_condition_to_string): Likewise.
33656 (add_condition_to_rtx): Likewise.
33657 (apply_attribute_uses): Likewise.
33658 (add_current_iterators): Likewise.
33659 (apply_iterators): Likewise.
33660 (initialize_iterators): Guard usage of apply_subst_iterator with
33661 #ifdef GENERATOR_FILE.
33662 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33663 (md_reader::read_mapping): Likewise.
33664 (add_define_attr_for_define_subst): Likewise.
33665 (add_define_subst_attr): Likewise.
33666 (read_subst_mapping): Likewise.
33667 (check_code_iterator): Likewise.
33668 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33670 (one_time_initialization): New function.
33671 (rtx_reader::read_until): New method.
33672 (read_flags): New function.
33673 (parse_reg_note_name): New function.
33674 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33675 Handle reuse_rtx ids.
33676 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33677 Add parsing support for RTL dumps, mirroring the special-cases in
33678 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33679 values, and calling handle_any_trailing_information.
33680 (rtx_reader::read_rtx_operand): Convert return type from void
33681 to rtx, returning return_rtx. Handle case 'e'. Call
33682 finalize_string on XSTR and XTMPL fields.
33683 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33684 "(nil)" values were omitted. Call the postprocess vfunc on the
33686 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33687 class ctor. Initialize m_in_call_function_usage. Call
33688 one_time_initialization.
33689 * rtl-tests.c (selftest::test_uncond_jump): Call
33690 set_new_first_and_last_insn.
33691 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33692 * selftest-rtl.c: New file.
33693 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33694 (selftest::get_insn_by_uid): New decl.
33695 * selftest-run-tests.c (selftest::run_tests): Call
33696 read_rtl_function_c_tests.
33697 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33698 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33701 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33703 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33704 operands in a special way. Assert that pos+len <= mode precision.
33706 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33708 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33709 3 argument Alias with unlimited for the negative form.
33710 (fno-vect-cost-model): Removed.
33712 2017-01-05 Martin Liska <mliska@suse.cz>
33714 * hsa-gen.c (gen_hsa_divmod): New function.
33715 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33717 2017-01-05 Martin Liska <mliska@suse.cz>
33720 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33723 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33725 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33726 small constant length operands.
33728 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33730 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33731 between loop iterations.
33733 2017-01-05 Martin Liska <mliska@suse.cz>
33736 * gimplify.c (gimplify_decl_expr): Compare to
33737 asan_poisoned_variables instread of checking flags.
33738 (gimplify_target_expr): Likewise.
33739 (gimplify_expr): Likewise.
33740 (gimplify_function_tree): Conditionally initialize
33741 asan_poisoned_variables.
33743 2017-01-04 Jeff Law <law@redhat.com>
33745 PR tree-optimizatin/78812
33746 * rtl.h (contains_mem_rtx_p): Prototype.
33747 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33748 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33749 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33750 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33751 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33753 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33755 * input.c (assert_char_at_range): Default-initialize actual_range.
33757 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33759 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33760 to match the caller.
33762 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33764 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33765 insns after final jump in test to emit dummy move.
33767 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33769 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33770 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33772 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33774 * multiple_target.c (create_dispatcher_calls): Init e_next.
33775 * tree-ssa-loop-split.c (split_loop): Init border.
33776 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33779 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33784 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33785 (altivec_register_operand): Do not return true if the operand
33786 contains a SUBREG mixing SImode and SFmode.
33787 (vsx_register_operand): Likewise.
33788 (vsx_reg_sfsubreg_ok): New predicate.
33789 (vfloat_operand): Do not return true if the operand contains a
33790 SUBREG mixing SImode and SFmode.
33791 (vint_operand): Likewise.
33792 (vlogical_operand): Likewise.
33793 (gpc_reg_operand): Likewise.
33794 (int_reg_operand): Likewise.
33795 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33796 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33797 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33799 (rs6000_emit_move_si_sf_subreg): New helper function.
33800 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33801 fixup SUBREGs involving SImode and SFmode.
33802 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33803 numbers for the new peephole2 optimization.
33804 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33805 the GLIBC math library that do AND/IOR/XOR operations on single
33806 precision floating point.
33807 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33808 target macros to say whether we need to avoid SUBREGs mixing
33810 (TARGET_ALLOW_SF_SUBREG): Likewise.
33811 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33812 (UNSPEC_SI_FROM_SF): Likewise.
33813 (iorxor): Change spacing.
33814 (and_ior_xor): New iterator for AND, IOR, and XOR.
33815 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33816 (movdi_from_sf_zero_ext): Likewise.
33817 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33818 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33819 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33820 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33821 (fms<mode>4): Likewise.
33822 (fnma<mode>4): Likewise.
33823 (fnms<mode>4): Likewise.
33824 (nfma<mode>4): Likewise.
33825 (nfms<mode>4): Likewise.
33827 2017-01-04 Marek Polacek <polacek@redhat.com>
33830 * doc/invoke.texi: Document -Wpointer-compare.
33832 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33834 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33837 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33838 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33841 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33844 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33845 documentation of the powerpc_popcntb_ok attribute.
33846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33847 code to issue warning messages if a requested CPU configuration is
33848 not supported by the binary (assembler and loader) toolchain.
33849 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33850 made to define a built-in function that has been disabled.
33851 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33852 made to define a built-in function that has been disabled.
33853 (altivec_init_builtins): Add comment explaining why definition
33854 of the DST built-in functions is not preceded by an assertion
33855 check. Add assertions to prevent ICE if attempts are made to
33856 define an altivec predicate or an abs* built-in function that has
33858 (htm_init_builtins): Add comment explaining why definition of the
33859 htm built-in functions is not preceded by an assertion check.
33861 2017-01-04 Jeff Law <law@redhat.com>
33863 PR tree-optimizatin/67955
33864 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33865 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33866 the points-to solution does not include pt_null. Use DECL_PT_UID
33869 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33871 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33872 Use gen_int_mode instead of gen_lopwart for const_int operands.
33874 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33876 PR tree-optimization/71563
33877 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33878 out of range value - has low bits known to be zero.
33880 2017-01-04 Alan Modra <amodra@gmail.com>
33882 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33883 * configure: Regenerate.
33884 * config.in: Regenerate.
33886 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33889 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33890 a substring of the message, but strcmp with the whole message. Ifdef
33891 ENABLE_NLS, translate the message first using dgettext.
33893 2017-01-03 Jeff Law <law@redhat.com>
33895 PR tree-optimizatin/78856
33896 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33897 (mark_threaded_blocks): Remove code to truncate thread paths that
33898 cross multiple loop headers. Instead invalidate the cached loop
33899 iteration information and handle case of a thread path walking
33900 into an irreducible region.
33902 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33905 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33906 assertions. Add support for doing the signbit if the IEEE 128-bit
33907 floating point value is in a GPR.
33908 * config/rs6000/rs6000.md (Fsignbit): Delete.
33909 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33910 Update the length attribute if the value is in a GPR.
33911 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33912 the sign or zero extension instruction, since the value is always 0/1.
33913 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33916 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33917 extracting SImode to a GPR register so that we can generate a
33918 store, limit the vector to be in a traditional Altivec register
33919 for the vextuwrx instruction.
33921 2017-01-03 Ian Lance Taylor <iant@google.com>
33923 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33925 2017-01-03 Martin Sebor <msebor@redhat.com>
33927 PR tree-optimization/78696
33928 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33929 precision. Use MPFR for %f for greater fidelity. Correct handling
33931 (pass_sprintf_length::compute_format_length): Set width and precision
33932 specified by asrerisk to void_node for vararg functions.
33933 (try_substitute_return_value): Adjust dump output.
33935 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33937 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33939 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33941 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33942 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33943 -mlra/-mno-lra was passed to the compiler.
33945 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33947 PR rtl-optimization/65618
33948 * emit-rtl.c (try_split): Move initialization of "before" and
33949 "after" to just before the call to emit_insn_after_setloc.
33951 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33953 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33955 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33957 * dwarf2out.c (gen_enumeration_type_die): When
33958 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33960 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33962 PR tree-optimization/78965
33963 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33964 Change first argument from const call_info & to call_info &. For %n
33965 set info.nowrite to false.
33967 PR middle-end/78901
33968 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33969 possibly throwing calls.
33971 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33972 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33973 and fns handling, rather than in a separate case SSA_NAME.
33975 2017-01-02 Jeff Law <law@redhat.com>
33977 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33978 fixes for first_period and second_period variables.
33980 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33983 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33984 (*insvqi_1): New insn pattern.
33985 (*insvqi_1_mem_rex64): Ditto.
33986 (*insvqi_2): Ditto.
33987 (*insvqi_3): Rename from *insvqi.
33989 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33991 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33993 * doc/cfg.texi (Edges): Remove reference to Java.
33994 (Maintaining the CFG): Ditto.
33996 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33998 PR middle-end/77674
33999 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34000 transparent aliases.
34002 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34004 PR middle-end/77484
34005 * predict.def (PRED_CALL): Update hitrate.
34006 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34007 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34008 into direct/indirect/polymorphic variants.
34010 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34012 Update copyright years.
34014 * gcc.c (process_command): Update copyright notice dates.
34015 * gcov-dump.c (print_version): Ditto.
34016 * gcov.c (print_version): Ditto.
34017 * gcov-tool.c (print_version): Ditto.
34018 * gengtype.c (create_file): Ditto.
34019 * doc/cpp.texi: Bump @copying's copyright year.
34020 * doc/cppinternals.texi: Ditto.
34021 * doc/gcc.texi: Ditto.
34022 * doc/gccint.texi: Ditto.
34023 * doc/gcov.texi: Ditto.
34024 * doc/install.texi: Ditto.
34025 * doc/invoke.texi: Ditto.
34027 Copyright (C) 2017 Free Software Foundation, Inc.
34029 Copying and distribution of this file, with or without modification,
34030 are permitted in any medium without royalty provided the copyright
34031 notice and this notice are preserved.